Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 06/24/2025 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-9 and 11-20 are rejected under 35 U.S.C. 101 because the claims recite an abstract idea without significantly more.
Regarding claim 1
Claim 1 – Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a process.
Claim 1 – Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
“using the trained key blank identification machine learning model, identifying one or more key blank part numbers for one or more key blanks that match the cut key;” – this limitation encompasses observing or evaluating visual characteristics of a cut key, comparing those characteristics with visual characteristics of a cut key, comparing those characteristics with visual characteristics or one or more candidate key blanks, and determining the part number of a key blank considered to match the cut key. The specification confirms that such identification was previously performed by a person paging though images of key blanks in printed manufacturing books and determining which key blank matched the cut key. See specification ¶ [0004]. Claim 1 does not require comparison against every key blank available in the market, processing of every visual data point, a minimum database size, a particular number of candidate blanks, or a specified speed or accuracy. The limitation therefore encompasses a person visually comparing the cut key with a limited number of key blanks and identifying the corresponding part number. The limitation falls within the mental-process grouping of abstract ideas because it recites an observation, evaluation, and judgment that can practically be performed in the human mind. Reciting that the evaluation is performed “using the trained key blank identification machine learning model” does not change the underlying character of the broadly recited identification, because the claim does not specify how the model performs the matching determination. See MPEP § 2106.04(a)(2)(III).
Claim 1 – Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No. There are no additional elements that integrate the judicial exception into a practical application. The additional elements:
“providing a user with a user interface to a key blank identification system;” – this limitation recites a user interface only at a high level of generality. The claim does not require a particular interface structure, interface-control technique, image-capture arrangement, or improvement in human-computer interaction. The interface merely provides access to the environment in which the mental comparison is implemented. This generally links the exception to a computer environment and does not meaningfully limit the exception. See MPEP §§ 2106.05(f) and 2106.05(h).
“receiving one or more images of a cut key from the user through the user interface of the key blank identification system;” – this limitation gathers the image information upon which the recited evaluation is performed. The limitation does not recite a particular image-capture, normalization, enhancement, segmentation, orientation-correction, or preprocessing operation. Receiving the images therefore constitutes insignificant pre-solution data-gathering activity. See MPEP § 2106.05(g).
“providing the one or more images of the cut key to a trained key blank identification machine learning model;” – this limitation transfers the gathered image data to the tool used to perform the recited identification. The limitation does not specify a particular data representation, model input structure, feature-extraction process, or technical interaction between the images and the model. It therefore amounts to incidental data transfer and an instruction to implement the recited evaluation in a machine-learning environment. See MPEP §§ 2106.05(f), 2106.05(g), and 2106.05(h).
“based on the identified one or more key blank part numbers, retrieving one or more images of the one or more matching key blanks from a key image database;” – this limitation retrieves information corresponding to the result of the mental evaluation after the matching part number has been identified. The limitation does not change or improve the manner in which the matching determination is performed. It therefore constitutes insignificant post-solution activity. See MPEP § 2106.05(g).
“displaying the one or more images of the one or more matching key blanks and the corresponding matched key blank part numbers to the user through the user interface of the key blank identification system;” – this limitation communicates the result of the mental evaluation to the user. The limitation does not require a particular display architecture or improve display technology. Merely displaying the result after the evaluation has been completed constitutes insignificant post-solution activity. See MPEP § 2106.05(g).
“and providing the user with one or more mechanisms to obtain one or more physical key blanks corresponding to one or more of the matched key blank part numbers.” – this limitation does not require that a physical key blank actually be obtained, cut, altered, or transformed. Under the broadest reasonable interpretation consistent with the specification, the mechanism may include a purchase link or information enabling the user to retrieve a blank from physical inventory. See specification ¶¶ [0105], [0112], and [0165]-[0166]. Providing such a mechanism merely applies or communicates the result after the matching determination. Because no physical key blank is required to be transformed into a duplicate key, this limitation does not effect a particular transformation and constitutes insignificant post-solution activity. See MPEP §§ 2106.05(c) and 2106.05(g).
Claim 1 – Step 2B – Does the claim recite additional elements that amount to significantly more than the judicial exception?
No. There are no additional elements that amount to significantly more than the judicial exception. The additional elements are:
“providing a user with a user interface to a key blank identification system;” – providing a generic user interface is a well-understood, routine, and conventional (WURC) use of computer technology. The specification states that the application may use a graphical, audio-based, touch-based, or any other known or later-developed user interface, indicating that no specialized interface is required. See specification ¶[0084]. See MPEP § 2106.05(d).
“receiving one or more images of a cut key from the user through the user interface of the key blank identification system;” – receiving digital images data through a generic computer interface is a well-understood, routine, and conventional (WURC) computer function recited at a high level of generality. The limitation does not require an unconventional image-acquisition or preprocessing operation. See MPEP §§ 2106.05(d) and 2106.05(g).
“providing the one or more images of the cut key to a trained key blank identification machine learning model;” – supplying input data to a trained computer model is a well-understood, routine, and conventional (WURC) use of a computer model according to its ordinary function. The specification states that a variety of machine-learning models and algorithms known to those of skill in the art may be used and identifies Siamese neural networks and convolutional neural networks as commonly used image-similarity models. See specification ¶[0064]. The limitation therefore does not recite an unconventional model architecture or data-input technique. See MPEP §§ 2106.05(d) and 2106.05(f).
“based on the identified one or more key blank part numbers, retrieving one or more images of the one or more matching key blanks from a key image database;” – retrieving a stored database record using an associated identifier is a well-understood, routine, and conventional (WURC) database function. The claim does not recite an unconventional database structure, indexing technique, retrieval algorithm, or storage arrangement. See MPEP §§ 2106.05(d) and 2106.05(g).
“displaying the one or more images of the one or more matching key blanks and the corresponding matched key blank part numbers to the user through the user interface of the key blank identification system;” – displaying retrieved images and associated textual information through a generic interface is a well-understood, routine, and conventional (WURC) computer-output function. The claim does not recite a specialized display technique. See MPEP §§ 2106.05(d) and 2106.05(g).
“and providing the user with one or more mechanisms to obtain one or more physical key blanks corresponding to one or more of the matched key blank part numbers.” – providing a purchase link, inventory-location mechanism, or similar option based on an identified product is well-understood, routine, and conventional (WURC) post-solution activity. The limitation does not require an unconventional transaction mechanism or physical transformation. See MPEP §§ 2106.05(c), 2106.05(d), and 2106.05(g).
Regarding claim 2
Claim 2 – Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a process.
Claim 2 – Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
Claim 2 – Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No. There are no additional elements that integrate the judicial exception into a practical application. The additional elements:
“wherein the key blank identification system includes an application provided to the user on a mobile device.” – this limitation merely restricts the computer environment in which the recited key-blank identification is performed. The limitation does not recite a particular mobile-device architecture, a specialized mobile application, a modified image-capture component, or an improvement to the operation of the mobile device. Rather, the limitation generally links the judicial exception to a mobile-device technological environment. Generally limiting a judicial exception to a particular technological environment or field of use odes not integrate the exception into a practical application. See MPEP §§ 2106.05(h) and 2106.05(a).
Claim 2 – Step 2B – Does the claim recite additional elements that amount to significantly more than the judicial exception?
No. There are no additional elements that amount to significantly more than the judicial exception. The additional elements are:
“wherein the key blank identification system includes an application provided to the user on a mobile device.” – providing a computer application through a generic mobile device constitutes use of well-understood, routine, and conventional (WURC) computer technology according to its ordinary function. The limitation does not recite an unconventional mobile-device component, application architecture, or arrangement of components. The specification’s disclosure that the same application may be provided through numerous interchangeable, generally available computing platforms further demonstrates that the mobile device is not used in an unconventional manner. See specification ¶¶ [0083]-[0084]. See MPEP §§ 2106.05(d) and 2106.05(h).
Regarding claim 3
Claim 3 – Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a process.
Claim 3 – Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
Claim 3 – Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No. There are no additional elements that integrate the judicial exception into a practical application. The additional elements:
“wherein the one or more images of the cut key received from the user include: an image of a first surface of the cut key; and an image of a second surface of the cut key.” – this limitation merely specifies that the information gathered for use in the recited key-blank identification includes images of two surfaces of the cut key. The limitation does not recite how the two images are aligned, normalized, combined, compared, or otherwise technically processed by the machine-learning model. Nor does the limitation require that different features be extracted from the respective surface images or that the use of two images changes the operation of the computer or machine-learning model. Accordingly, obtaining an image of each surface merely supplies additional input data upon which the recited evaluation is performed and constitutes insignificant pre-solution data-gathering activity. See MPEP § 2106.05(g). The additional data does not, by itself, reflect an improvement to image-processing or machine-learning technology. See MPEP § 2106.05(a).
Claim 3 – Step 2B – Does the claim recite additional elements that amount to significantly more than the judicial exception?
No. There are no additional elements that amount to significantly more than the judicial exception. The additional elements are:
“wherein the one or more images of the cut key received from the user include: an image of a first surface of the cut key; and an image of a second surface of the cut key.” – retrieving two digital images rather than one constitutes use of ordinary image-acquisition and data-input functionality according to its conventional function. The specification describes promoting the user to photograph the first and second surfaces using an image-capture button or to upload existing photographs, without identifying a specialized camera arrangement, image format, or unconventional acquisition technique. See specification ¶¶ [0091]-[0094] and [0139]-[0143]. Thus, receiving images of both surfaces is well-understood, routine, and conventional (WURC) data-gathering activity recited at a high level of generality. See MPEP §§ 2106.05(d) and 2106.05(g).
Regarding claim 4
Claim 4 – Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a process.
Claim 4 – Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
“wherein the trained key blank identification machine learning model utilizes one or more image similarity algorithms to identify similarities between images of cut keys and images of key blanks,”
Claim 4 – Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No. There are no additional elements that integrate the judicial exception into a practical application. The additional elements:
“and further wherein the one or more similarity algorithms are selected from the group of image similarity algorithms consisting of: Siamese Neural Networks; and Convolutional Neural Networks.” - this limitation merely restricts the technological tool used to perform the recited image-similarity evaluation to one or more named types of neural networks. The limitation does not recite a particular Siamese neural-network or convolutional neural-network architecture, arrangement of layers, embedding structure, feature map, loss function, distance metric, training procedure, or other technical operation by which image similarity is determined. The specification describes Siamese neural networks and convolutional neural networks as examples of algorithms “commonly used for image similarity detection”, rather than as specially configured or technologically improved neural networks. See specification ¶ [0064]. Thus, the limitation merely instructs that the recited image-similarity evaluation be implemented using a selected, generally identified neural-network tool. Restricting performance of a judicial exception to a particular type of computer tool, without reciting a particular technological improvement or manner of operation, does not integrate the exception into a practical application. See MPEP §§ 2106.05(a), 2106.05(f), and 2106.05(h).
Claim 4 – Step 2B – Does the claim recite additional elements that amount to significantly more than the judicial exception?
No. There are no additional elements that amount to significantly more than the judicial exception. The additional elements are:
“and further wherein the one or more similarity algorithms are selected from the group of image similarity algorithms consisting of: Siamese Neural Networks; and Convolutional Neural Networks.” – using a Siamese neural network or convolutional neural network for image-similarity detection constitutes use of known machine-learning tools according to their ordinary functions. The specification expressly characterizes these neural networks as examples of algorithms commonly used for image-similarity detection and does not identify an unconventional architecture or operation required by claim 4. See specification ¶ [0064]. Accordingly, the limitation recites well-understood, routine, and conventional (WURC) machine-learning technology at a high level of generality and does not supply an inventive concept. See MPEP § 2106.05(d).
Regarding claim 5
Claim 5 – Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a process.
Claim 5 – Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
“wherein the trained key blank identification machine learning model utilizes one or more weighting parameters to rank potential key blank matches,” – this limitation recites assigning relative importance to image-similarity information and using the weighted information to order candidate key blanks according to their likelihood of matching the cut key. Applying weights to information and ranking candidate based on the weighted information describes a mathematical relationship or calculation. The limitation also encompasses evaluating the relative significance of visual similarities and ordering the candidate matches, which are evaluations and judgments that can practically be performed in the human mind, with or without the aid of pen and paper. See MPEP §§ 2106.04(a)(2)(I) and 2106.04(a)(2)(III).
“and further wherein similarities between portions of key images depicting the key bottom are weighted more heavily than similarities between portions of key images depicting the key head.” – this limitation expressly recites a relative weighting relationship in which similarities associated with one image region are assigned greater significance than similarities associated with another image region. The relative weighting of key-bottom and key-head similarities describes a mathematical relationship used in evaluating the candidate key blanks. The limitation also encompasses a person determining that similarities in the key bottom are more important than similarities in the key head when judging whether a candidate key blank matches the cut key. See MPEP §§ 2106.04(a)(2)(I) and 2106.04(a)(2)(III).
Claim 5 – Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No. There are no additional elements that integrate the judicial exception into a practical application. The additional elements:
Claim 5 – Step 2B – Does the claim recite additional elements that amount to significantly more than the judicial exception?
No. There are no additional elements that amount to significantly more than the judicial exception. The additional elements are:
Regarding claim 6
Claim 6 – Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a process.
Claim 6 – Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
“wherein similarities between portions of key images depicting the key bottom are weighted at 90% and similarities between portions of key images depicting the key head are weighted at 10%.” – this limitation expressly assigns numerical weights to two categories of image-similarity information and uses a 90%/10% relationship to determine the relative significance of the key-bottom and key-head similarities. Assigning percentage values to respective image regions and applying those percentages when evaluating similarity recites a mathematical relationship and mathematical calculation. The limitation also encompasses evaluating candidate key blanks by giving substantially greater importance to similarities in the key bottom than to similarities in the key head. Such relative evaluation and judgment can practically be performed by a person, including with the aid of pen and paper. See MPEP §§ 2106.04(a)(2)(I) and 2106.04(a)(2)(III).
Claim 6 – Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No. There are no additional elements that integrate the judicial exception into a practical application. The additional elements:
Claim 6 – Step 2B – Does the claim recite additional elements that amount to significantly more than the judicial exception?
No. There are no additional elements that amount to significantly more than the judicial exception. The additional elements are:
Regarding claim 7
Claim 7 – Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a process.
Claim 7 – Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
“confirmation as to whether one of the matched key blank part numbers is the correct key blank part number is requested from the user;” – this limitation recites asking the user to evaluate the displaying candidate key blank part numbers and determine whether one of the candidates is correct. Determining whether a candidate part number correctly corresponds to the cut key is an evaluation and judgment that can practically be performed in the human mind. See MPEP § 2106.04(a)(2)(III).
“and confirmation that one of the matched key blank part numbers is the correct key blank part number is received from the user.” – this limitation recites obtaining the result of the user’s evaluation and judgment that a particular candidate part number correctly identifies the matching key blank. The confirmation represents a conclusion reached through observation, evaluation, and judgment that can practically be performed in the human mind. See MPEP § 2106.04(a)(2)(III).
Claim 7 – Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No. There are no additional elements that integrate the judicial exception into a practical application. The additional elements:
Claim 7 – Step 2B – Does the claim recite additional elements that amount to significantly more than the judicial exception?
No. There are no additional elements that amount to significantly more than the judicial exception. The additional elements are:
Regarding claim 8
Claim 8 – Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a process.
Claim 8 – Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
“wherein upon receiving confirmation that one of the matched key blank part numbers is the correct key blank part number, one or more images in the key image database are annotated with the correct key blank part number.” – this limitation recites associating one or more key images with a particular key blank part number based on the user’s confirmation that the part number is correct. Annotating an image with a corresponding part number constitutes labeling or classifying information according to the result of an evaluation. A person can practically perform this operation by reviewing the confirmed result and writing, recording, or otherwise associating the correct part number with the corresponding key image. The recitation that the images are contained in a key image database merely identifies the information being labeled and does not change the underlying character classification.
Claim 8 – Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No. There are no additional elements that integrate the judicial exception into a practical application. The additional elements:
Claim 8 – Step 2B – Does the claim recite additional elements that amount to significantly more than the judicial exception?
No. There are no additional elements that amount to significantly more than the judicial exception. The additional elements are:
Regarding claim 9
Claim 9 – Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a process.
Claim 9 – Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
Claim 9 – Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No. There are no additional elements that integrate the judicial exception into a practical application. The additional elements:
“wherein upon receiving confirmation that one of the matched key blank part numbers is the correct key blank part number, the user is provided with one or more mechanisms to obtain one or more physical key blanks corresponding to the correct key blank part number.” – this limitation provides an option or mechanism based on the result of the abstract evaluation after the user has confirmed the correct key-blank part number. This limitation does not require that a physical key blank actually be obtained, delivered, cut, modified, or otherwise transformed. Under the broadest reasonable interpretation consistent with the specification, the recited mechanism may include presenting a purchase link or providing the correct part number so that the user can locate the corresponding key blank in physical inventory. See specification ¶¶ [0105], [0112], and [0165]-[0166]. Thus, the limitation merely uses the result of the abstract key-identification process to provide the user with a subsequent opportunity to obtain a corresponding product. Because the limitation occurs after the matching determination and does not affect how the correct key-blank part number is identified, it constitutes insignificant post-solution activity. See MPEP § 2106.05(g). Because the claim does not require acquisition or physical modification of the blank, the limitation also does not effect a transformation of a particular article into a different state or thing. See MPEP § 2106.05(c).
Claim 9 – Step 2B – Does the claim recite additional elements that amount to significantly more than the judicial exception?
No. There are no additional elements that amount to significantly more than the judicial exception. The additional elements are:
“wherein upon receiving confirmation that one of the matched key blank part numbers is the correct key blank part number, the user is provided with one or more mechanisms to obtain one or more physical key blanks corresponding to the correct key blank part number.” – providing a purchase link, inventory-location option, or similar mechanism based on an identified product constitutes routine post-solution activity performed according to its ordinary function. The claim does not recite an unconventional purchasing mechanism, inventory system, fulfillment process, or interaction between the computer system and the physical key blank. The specification describes the mechanism at a high level of generality, including providing a link enabling the user to purchase they key blank or providing the part number to that the user may locate the blank in a physical storage location. See specification ¶¶ [0112], [0115], and [0165]-[0166]. Accordingly, the limitation recites well-understood, routine, and conventional (WURC) post-solution activity and does not provide an inventive concept. See MPEP §§ 2106.05(d) and 2106.05(g).
Regarding claim 11
Claim 11 – Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a process.
Claim 11 – Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
“using the trained key blank identification machine learning model, identifying one or more key blank part numbers for one or more key blanks that match the particular cut key;” – this limitation encompasses observing or evaluating visual characteristics of a cut key, comparing those characteristics with visual characteristics of a cut key, comparing those characteristics with visual characteristics or one or more candidate key blanks, and determining the part number of a key blank considered to match the cut key. The specification confirms that such identification was previously performed by a person paging though images of key blanks in printed manufacturing books and determining which key blank matched the cut key. See specification ¶ [0004]. Claim 1 does not require comparison against every key blank available in the market, processing of every visual data point, a minimum database size, a particular number of candidate blanks, or a specified speed or accuracy. The limitation therefore encompasses a person visually comparing the cut key with a limited number of key blanks and identifying the corresponding part number. The limitation falls within the mental-process grouping of abstract ideas because it recites an observation, evaluation, and judgment that can practically be performed in the human mind. Reciting that the evaluation is performed “using the trained key blank identification machine learning model” does not change the underlying character of the broadly recited identification, because the claim does not specify how the model performs the matching determination. See MPEP § 2106.04(a)(2)(III).
“upon user confirmation of identification of a correct key blank, annotating the one or more images of the particular cut key with the key blank part number corresponding to the correct key blank;” – this limitation recites assigning or associating the confirmed key-blank part number with the images of the particular cut key. It therefore encompasses labeling and classifying the images according to the user’s judgment that a particular key blank is correct. A person can practically perform this operation by determining the correct part number and writing or recording that part number in association with the corresponding key images. See MPEP § 2106.04(a)(2)(III).
“and utilizing the refined key blank identification machine learning model to identify unique key blank part numbers that correspond with a plurality of cut keys.” – this limitation encompasses evaluating a plurality of cut keys, determining which key blanks correspond to the cut keys, and identifying the associated part numbers. The claim does not specify how the refined model extracts image features, compares representations, calculates similarity, or technically determines correspondence. The limitation therefore broadly recites the same type of visual comparison and judgment identified above, merely applied to a plurality of cut keys. See MPEP § 2106.05(a)(2)(III).
Claim 11 – Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No. There are no additional elements that integrate the judicial exception into a practical application. The additional elements:
“aggregating a plurality of images of one or more cut keys;” – this limitation collects the image information used to create training data and perform the claimed identification. The limitation does not recite a particular image-acquisition device, image-processing operation, normalization technique, or data structure. See MPEP § 2106.05(g).
“aggregating a plurality of images of one or more key blanks, wherein each of the key blanks correspond to at least one of the one or more cut keys;” – this limitation similarly gathers candidate key-blank images used in preparing the model and performing the identification. It does not recite a particular technical process for establishing, representing, or storing the claimed correspondence. The limitation therefore constitutes insignificant data-gathering activity. See MPEP § 2106.05(g).
“generating key blank identification training data based on the plurality of images of one or more cut keys and the plurality of images of one or more key blanks;” – this limitation recites the result of generating training data but does not define how the images are formatted, normalized, paired, portioned, encoded, or technically transformed into the training data. The limitation therefore invokes generic data preparation as an antecedent to performing the abstract identification rather than reciting a particular technological solution. See MPEP §§ 2106.05(f) and 2106.05(g).
“generating a trained key blank identification machine learning model by training one or more machine learning models to identify key blanks using the key blank identification training data;” – this limitation recites training at a result-oriented level. The claim does not recite a particular model architecture, loss function, selected training algorithm, feature-extraction process, embedding structure, distance metric, validation technique, method of modifying model parameters, or another technical training operation. Therefore, the limitation uses generic machine-learning training as the tool for automating the abstract comparison and identification process. Merely invoking a computer or machine-learning tool to perform an existing evaluative process does not integrate the exception into a practical application. See MPEP §§ 2106.05(a) and 2106.05(f).
“receiving, by the trained key blank identification machine learning model, one or more images of a particular cut key from a user through a user interface of a key blank identification system;” – this limitation supplies the image data upon which the abstract evaluation is performed. The limitation does not recite a particular image-capture technique, image alignment process, orientation correction, segmentation operation, or preprocessing technique. Accordingly, the limitation is insignificant data-gathering activity and merely links the abstract evaluation to a generic user-interface environment. See MPEP §§ 2106.05(g) and 2106.05(h).
“based on the identified one or more key blank part numbers, retrieving one or more images of the one or more matching key blanks from a key image database;” – this limitation retrieves information corresponding to the result of the abstract identification. It does not affect how the identification is performed and therefore constitutes insignificant post-solution activity. See MPEP § 2106.05(g).
“displaying the one or more images of the one or more matching key blanks and the corresponding key blank part numbers to the user through a user interface of the key blank identification system;” – this limitation merely communicates the result of the abstract evaluation. It does not recite an improvement to display technology or a specialized manner of presenting the result. Accordingly, it constitutes insignificant pre-solution activity. See MPEP § 2106.05(g).
“receiving user feedback from the user through the user interface of the key blank identification system;” – this limitation gathers the user’s response concerning the displayed result. It does not recite a particular feedback-validation mechanism, error-control process, confidence calculation, or other technical operation. It therefore constitutes insignificant data gathering and generally links the claimed process to a user-interface environment. See MPEP §§ 2106.05(g) and 2106.05(h).
“incorporating the user feedback into the key blank identification training data to generate a refined trained key blank identification machine learning model;” – this limitation states the desired result of refining the model based on feedback but does not recite how the feedback changes the model. The claim does not specify whether the model is retrained, fine-tuned, incrementally updated, or otherwise modified. Nor does it identify the model parameters changed, the training algorithm used, an error-correction procedure, or a validation criterion.
Claim 11 – Step 2B – Does the claim recite additional elements that amount to significantly more than the judicial exception?
No. There are no additional elements that amount to significantly more than the judicial exception. The additional elements are:
“aggregating a plurality of images of one or more cut keys;” – collecting digital images is a well-understood, routine, and conventional (WURC) data-acquisition function performed according to its ordinary use. See MPEP §§ 2106.05(d) and 2106.05(g).
“aggregating a plurality of images of one or more key blanks, wherein each of the key blanks correspond to at least one of the one or more cut keys;” – collecting and organizing candidate images according to associated subject matter is well-understood, routine, and conventional (WURC) data preparation and does not recite an unconventional image-acquisition or data-management technique. See MPEP §§ 2106.05(d) and 2106.05(g).
“generating key blank identification training data based on the plurality of images of one or more cut keys and the plurality of images of one or more key blanks;” – preparing image data for use as training data is recited only at a high level of generality and is a well-understood, routine, and conventional (WURC) preprocessing step in machine learning model training. It does not require an unconventional formatting, labeling, partitioning, encoding, or preprocessing operation. It therefore does not provide an inventive concept. See MPEP § 2106.05(d).
“generating a trained key blank identification machine learning model by training one or more machine learning models to identify key blanks using the key blank identification training data;” – training a generic machine-learning model with training data according to its ordinary function is well-understood, routine, and conventional (WURC). The specification states that known models and algorithms may be used and identifies commonly used image-similarity models. See specification ¶¶ [0063]-[0064] and [0178]-[0179]. See MPEP §§ 2106.05(d) and MPEP § 2106.05(f).
“receiving, by the trained key blank identification machine learning model, one or more images of a particular cut key from a user through a user interface of a key blank identification system;” – receiving digital images through a generic interface and supplying those images as model input are well-understood, routine, and conventional (WURC) computer functions. The claim does not require unconventional image-input hardware or preprocessing. See MPEP §§ 2106.05(d) and 2106.05(g).
“based on the identified one or more key blank part numbers, retrieving one or more images of the one or more matching key blanks from a key image database;” – retrieving stored records based on an associated identifier is a routine database function. The limitation does not recite an unconventional database structure or retrieval mechanism. See MPEP §§ 2106.05(d) and 2106.05(g).
“displaying the one or more images of the one or more matching key blanks and the corresponding key blank part numbers to the user through a user interface of the key blank identification system;” – displaying retrieved images and associated text through a generic user interface is a well-understood, routine, and conventional (WURC) computer-output function. See MPEP § 2106.05(d).
“receiving user feedback from the user through the user interface of the key blank identification system;” – receiving user input through a generic interface is a well-understood, routine, and conventional (WURC) input function. The limitation does not recite an unconventional feedback-acquisition or verification process. See MPEP §§ 2106.05(d) and 2106.05(g).
“incorporating the user feedback into the key blank identification training data to generate a refined trained key blank identification machine learning model;” – adding newly labeled information to training data and using the updated data to refine a model are well-understood, routine, and conventional (WURC) machine-learning operations recited according to their ordinary functions. The claim does not recite an unconventional retraining, fine-tuning, or model-update technique. See MPEP §§ 2106.05(d) and 2106.05(f).
Regarding claims 12-19
Claims 12-19 depend, directly or indirectly, from independent claim 11 and incorporate the judicial exception identified therein. The additional limitations recited by claims 12-19 correspond to limitations previously analyzed with respect to claims 3-8 and merely further define the same abstract mental evaluation, mathematical relationships, and/or data-labeling operations performed during key blank identification and machine-learning training.
Under Step 2A, Prong One, the additional limitations of claims 12-19 recite the same abstract ideas discussed above for the corresponding method claims. Specifically:
claim 12 recites annotating images with a corresponding key blank part number, which constitutes classification and labeling of information based on an evaluative determination, consistent with the mental-process grouping of abstract ideas under MPEP § 2106.04(a)(2)(III);
claim 13 recites receiving images depicting first and second surfaces of a cut key, which merely gathers additional image information used in the abstract evaluation;
claim 14 limits the image-similarity algorithms to Siamese Neural Networks and/or Convolutional Neural Networks, while still performing the same abstract image-comparison operations identified above;
claims 15-17 recite weighting relationships and numerical weighting values used to rank potential key blank matches, which constitute mathematical relationships/calculations and evaluative determinations, consistent with MPEP § 2106.04(a)(2)(I) and MPEP § 2106.04(a)(2)(III);
claim 18 recites receiving a user’s indication regarding whether an identified key blank part number is correct, which constitutes an evaluative judgment; and
claim 19 recites annotating images based upon the user’s indication, which constitutes classification and labeling of information based upon the evaluative judgment.
Under Step 2A, Prong Two, the additional limitations of claims 12-19 do not integrate the judicial exception into a practical application because the claims do not recite a particular image-processing technique, specialized machine-learning architecture, unconventional model-training operation, specific image-segmentation process, defined feature-extraction technique, or other technological implementation beyond using generic computing components to perform the recited abstract operations. Instead, the additional limitations merely gather additional image data, further define the abstract mathematical evaluation, receive evaluative feedback, or store the results of that evaluation. Accordingly, the additional limitations amount to no more than insignificant extra-solution activity under MPEP § 2106.05(g), instructions to apply the abstract idea using generic computer technology under MPEP § 2106.05(f), and/or field-of-use limitations under MPEP § 2106.05(h), as discussed above for the analogous limitations of claims 3-8.
Under Step 2B, the additional limitations of claims 12-19 likewise do not provide an inventive concept. The additional image acquisition, generic neural-network selection, weighting relationships, user feedback, and annotation operations merely employ well-understood, routine, and conventional (WURC) computer functions and machine-learning techniques according to their ordinary functions. The claims do not recite an unconventional neural-network architecture, unconventional training methodology, unconventional weighting mechanism implemented by the computer, or other technological improvement sufficient to transform the judicial exception into patent-eligible subject matter. See MPEP § 2106.05(d).
Accordingly, because claims 12-19 merely further limit the judicial exception identified in claim 11 without integrating the exception into a practical application or reciting significantly more than the judicial exception, claims 12-19 are rejected under 35 U.S.C. § 101 for the reasons discussed above with respect to claim 11 and the corresponding analysis of claims 3-8, respectively.
Regarding claim 20
Claim 20 recites substantially the same limitations already analyzed in independent claim 11, recited in system form. The functional language associated with the claimed “key image database”, “one or more machine learning models”, “key blank identification application”, “at least one processor”, and “at least one memory” merely requires generic computing components configured to perform the same abstract operations already identified in method claim 11, including aggregating images of cut keys and key blanks, generating key blank identification training data, training one or more machine learning models, identifying corresponding key blank part numbers through image comparison, receiving user feedback, annotating images based upon the user’s evaluative determination, refining the trained machine learning model, and utilizing the refined model to identify corresponding key blank part numbers for additional cut keys. The system formulation does not materially change the eligibility analysis.
Under Step 2A, Prong One, the corresponding system limitations recite the same abstract ideas identified above for the corresponding method claim, including mental processes involving observation, comparison, evaluation, classification, and judgment, as well as mathematical concepts involving similarity analysis, weighting, and ranking operations where applicable, consistent with MPEP §§ 2106.04(a)(2)(I) and 2106.04(a)(2)(III).
Under Step 2A, Prong Two, the claimed system does not integrate the judicial exception into a practical application because the claim does not recite a particular image-processing technique, specialized image-acquisition hardware, unconventional machine-learning architecture, specific model-training algorithm, defined model-update technique, particular feature-extraction mechanism, or other technological implementation beyond generic computing components configured to perform the recited abstract operations. The addit0onal system components merely implement the abstract ideas using generic computer hardware and amount to no more than instructions to apply the abstract ideas using generic computer technology under MPEP § 2106.05(f), insignificant data gathering and post-solution activity under MPEP § 2106.05(g) and/or generally linking the abstract ideas to a technological environment under MPEP § 2106.05(h), as discussed above with respect to claim 11.
Under Step 2B, the claimed key image database, one or more machine learning models, key blank identification application, at least one processor, and at least one memory are generic computing components performing their ordinary and conventional functions and are well-understood, routine, and conventional (WURC) under MPEP § 2106.05(d). The specification describes the key image database as storing images and associated information, the key blank identification application as a generic application executable on conventional computing devices, and the machine learning models as known image-similarity models, including Siamese Neural Networks and Convolutional Neural Networks. Likewise, the claim does not recite an unconventional processor architecture, specialized database structure, unconventional machine-learning architecture, unconventional model-training methodology, unconventional model-refinement technique, or any other inventive concept sufficient to transform the judicial exception into patent-eligible subject matter.
Accordingly, because claim 20 merely implements the same abstract mathematical concepts and mental processes identified in claim 11 using generic computing components, without integrating the judicial exception into a practical application or reciting significantly more than the judicial exception itself, claim 20 is rejected under 35 U.S.C. § 101 for the same reasons discussed above with respect to claim 11.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-20 are rejected under 35 U.S.C. 103 as being unpatentable over Gregory Marsh (US20130173044A1) in view of Michael Chertok (US20160196479A1) and further in view of Michael A. Bass (US20040095380A1).
Regarding claim 1, teach a computing system implemented method for identifying key blank part numbers comprising:
“providing a user with a user interface to a key blank identification system;” – Marsh teaches this limitation. Marsh teaches a display and one or more input device, which may be combined as a touchscreen:
“touch sensitive display (or touchscreen device).” (Marsh, pg. 2, ¶[0032])
Thus, Marsh teaches providing the user with a user interface to the key-blank identification system.
“receiving one or more images of a cut key from the user through the user interface of the key blank identification system;” – Marsh teaches this limitation. Marsh teaches that the user may photograph the key and provide one or more images to a processor:
“photograph a key and supply one or more corresponding images” (Marsh, pg. 7, ¶[0090])
Marsh further teaches that an application may assist the user in capturing suitable key images:
“In some embodiments, users can scan a key on their smartphone or tablet computer. Furthermore, … an application can be downloaded to assist users in capturing quality images of their key. From a mobile application or website platform, users can also request a duplicate of their scanned key be sent via mail to an address of their choosing” (Marsh, pg. 7, ¶[0091])
Thus, Marsh teaches receiving one or more images of a cut key from the user through the user interface.
“providing the one or more images of the cut key ” – Marsh teaches this limitation in part. Marsh teaches providing the user-supplied key images to a hardware processor executing machine-vision software. The processor analyzes the images to detect the key’s bitting pattern and key-blank type. Marsh further teaches that machine-learning algorithms may be used with the machine-version software:
“Machine learning algorithms can also be employed to improve the accuracy and capabilities of the machine vision software.” (Marsh, pg. 7, ¶[0083])
Thus, Marsh teaches providing the cut-key images to machine-vision software that may employ machine learning, but does not expressly teach that the images are provided to a trained key blank identification machine learning model.
identifying one or more ” – Marsh teaches this limitation in part. Marsh teaches measuring width variations in the key image, comparing those variations with known key-blank data, and determining the corresponding key-blank type:
“These width variations can be compared to known key blank data to determine a key blank type (or milling type). This data can be used to select the appropriate blank type during the key replication.” (Marsh, pg. 7, ¶[0081])
Thus, Marsh teaches identifying a key blank that matches the cut key.
“displaying ” – Marsh teaches this limitation in part. Marsh teaches communicating key-type and bitting information to the user through a kiosk, mobile application, or website platform:
“In some embodiments, information on a key type and bitting information of a key provided by the user can be sent to the user … For example, the information can be sent to the user by e-mail, text message, mail, or any other suitable manner of sending the information.” (Marsh, pg. 5, ¶[0065])
“and providing the user with one or more mechanisms to obtain one or more physical key blanks corresponding to one or more of the matched ” – Marsh teaches this limitation in part. Marsh teaches that the user may determine the required blank type through a computer or mobile application, purchase the blank at a retail location, obtain the blank from an attendant, or request a duplicate through the application:
“For example, the user can look up the type of key blank using a computer application, mobile application, or web platform.” (Marsh, pg. 4, ¶[0053])
“In some cases a user can buy a blank of the required type at a retail location” (Marsh, pg. 4, ¶[0053])
“or acquire the appropriate blank from an attendant (e.g., a clerk).” (Marsh, pg. 4, ¶[0053])
“request a duplicate of their scanned key be sent via mail to an address of their choosing” (Marsh, pg. 7, ¶[0091])
Thus, Marsh provides mechanisms for obtaining a physical key blank corresponding to the identified blank type, but does not expressly identify the physical blank using a matched key blank part number.
Marsh does not teach these limitations and/or portions of:
“to a trained key blank identification machine learning model”
“using the trained key blank identification machine learning model … key blank part numbers for one or more”
“based on the identified one or more key blank part numbers, retrieving one or more images of the one or more matching key blanks from a key image database;”
“the one or more images of the one or more matching key blanks and the corresponding matched key blank part numbers”
“key blank part numbers.”
Chertok, however, teaches these limitations and/or portions of:
“to a trained key blank identification machine learning model” – Chertok teaches a convolutional neural network having parameters established during a training period and trained using labeled images:
“CNN … is trained by applying it to a training set or pre-loaded images” (Chertok, pg. 3, ¶[0020])
Chertok does not expressly describe the trained network as a key blank identification model. Marsh, however, already teaches applying machine-learning-enhanced machine vision to determine the key-blank type from key images. Thus, Marsh in view of Chertok teaches a trained key-blank-identification machine-learning model.
“using the trained key blank identification machine learning model” – Chertok teaches supplying a query image and a reference image to the trained neural network, generating descriptors for the images, comparing corresponding descriptors, and determining the similarity between the images:
“After determining the weights of each descriptor similarity (i.e., the weight of each layer of the network, and more generally each image descriptor), system 400 determines image similarity between a pair of images by performing the method steps of FIG. 2.” (Chertok, pg. 7, ¶[0069])
“Descriptor comparator 408 determines the descriptor similarity between the first image descriptor of the query image and the first image descriptor of the reference image, and so forth.” (Chertok, pg. 7, ¶[0070])
“Thereafter, image comparator 412 defines the image similarity between the query image and the reference image as a function of the weighted descriptor similarities. System 400 employs the determined image similarity between the query image and the reference image for performing various visual tasks, such as image retrieval or machine vision.” (Chertok, pg. 7, ¶[0071])
In the proposed combination, Marsh’s user-supplied cut-key image would constitute Chertok’s query image, and an image associated with a known key blank would constitute Chertok’s reference image. The trained network would therefore be used to identify the known key blank whose image most closely corresponds to the submitted cut-key image.
“” – Chertok teaches this limitation in part. Chertok teaches storing and retrieving query and reference images from data storage and applying the trained network to the stored images:
“retrieves a query image and a reference image from data storage” (Chertok, pg. 7, ¶[0069])
Chertok further teaches using the determined image similarity for image retrieval or machine-vision operations:
“System 400 employs the determined image similarity between the query image and the reference image for performing various visual tasks, such as image retrieval or machine vision.” (Chertok, pg. 7, ¶[0071])
Thus, Chertok teaches retrieving stored images and identifying or retrieving matching reference images using the trained image-similarity network.
Chertok nor Marsh teach these remaining limitations and/or portions of:
“key blank part numbers for one or more”
“based on the identified one or more key blank part numbers;”
“key blanks … key image database;”
“the one or more images of the one or more matching key blanks and the corresponding matched key blank part numbers”
“key blank part numbers.”
Bass, however, teaches these remaining limitations and/or portions of:
“key blank part numbers for one or more” – Bass teaches identifying a key blank by manufacturer or part number, including an ILCO number, ACE stock number, or UPC code:
“identification by manufacturer or part number” (Bass, pg. 5, ¶[0055])
Thus, Bass teaches identifying matching key blanks using corresponding key-blank part numbers.
“based on the identified one or more key blank part numbers;” and reference images depicting “key blanks” stored in a “key image database” – Bass teaches the remaining portions of this limitation in combination with Marsh and Chertok. Bass teaches a database algorithm to identify one or more appropriate key blanks and associating each identified key blank with both part-number information and a corresponding key-blank image:
“The information input into the system will enable; the algorithm to select the appropriate key-blank(s) for replication from the master key.” (Bass, pg. 5, ¶[0054])
Bass further states that the database may provide:
“identification by manufacturer or part number (for example, ILCO number, ACE stock number, UPC code and/or the Jaw Selection)” (Bass, pg. 5, ¶[0055])
Bass additionally teaches:
“The visual display (i.e., the monitor, CRT or video system) would show a picture of the identified blank in order to permit visual verification by the user” (Bass, pg. 7, ¶[0068])
Bass also describes a key-blank display arrangement comprising:
“pictures of specific key blanks, bar codes to assist in the identification, inventory and tracking of each key blank” (Bass, pg. 7, ¶[0072])
Thus, Bass teaches database records associating identified matching key blanks with corresponding key-blank part numbers and key-blank pictures. In the combined system, Chertok’s stored reference images would constitute Bass’s pictures of specific key blanks, and each reference image would be associated with Bass’s corresponding manufacturer, part number, or code-number record. Upon identification of a matching key-blank record by the Marsh-Chertok trained model, the associated key-blank image would be retrieved from the key-image database based on the identified key-blank part number. Accordingly, Marsh in view of Chertok and Bass teaches this limitation.
“the one or more images of the one or more matching key blanks and the corresponding matched key blank part numbers” – Bass teaches displaying a picture of the identified key blank to permit visual verification and displaying corresponding manufacturer or part-number information:
“show a picture of the identified blank” (Bass, pg. 7, ¶[0068])
Bass also teaches displaying the manufacturer or part number of the identified blank:
“All identified key-blanks can be highlighted by the display rack for easy retrieval by the, user.” (Bass, pg. 5, ¶[0054])
Thus, Bass teaches displaying an image of the identified matching key blank together with its corresponding key-blank part number:
“key blank part numbers.” – Bass teaches using the identified key-blank information to highlight the physical blank for retrieval or to cause an automated retrieval system to retrieve the identified blank. Bass teaches a retrieval unit configured to:
“automatically pick the key blank from a storage location.” (Bass, pg. 6, ¶[0063])
Thus, Bass teaches providing a mechanism for obtaining the physical key blank corresponding to the identified key-blank part number.
A person of ordinary skill in the art would have been motivated to associate the key-blank type identified by the Marsh-Chertok system with Bass’s key-blank part-number and picture records because Marsh already uses the identified blank type to select, retrieve, purchase, or produce the appropriate physical blank.
Bass teaches that displaying the corresponding blank picture and manufacturer or inventory part number facilitates visual verification and accurate retrieval. Adding Bass’s picture and part-number records would therefore have predictably reduced blank-selection errors and facilitated the purchasing, inventory-location, or automated retrieval operations already contemplated by Marsh.
Accordingly, Marsh teaches receiving user-supplied images of a cut key through a user interface, analyzing the images to identify a corresponding key-blank type, and providing mechanisms for obtaining the identified physical blank. Chertok teaches using a trained neural-network image-similarity model to compare a query image with stored reference images and retrieve matching images. Bass teaches associating key-blank images with corresponding key-blank part numbers, displaying the matching images and part numbers to the user, and facilitating retrieval of the identified physical blank.
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date to modify Marsh’s key-blank-identification system to employ Chertok’s trained image-similarity model and Bass’s key-blank-image and part-number database, thereby receiving a cut-key image and part number, and providing a mechanism to obtain the corresponding physical key blank.
Regarding claim 2, Marsh in view of Chertok and further in view of Bass teach the computing system implemented method of Claim 1
“wherein the key blank identification system includes an application provided to the user on a mobile device.” – Marsh teaches this limitation. Marsh teaches that a user may use a smartphone or tablet computer to capture images of a key and that an application is downloaded onto the mobile device to assist the user in capturing suitable key images:
“In such embodiments, for example, a user can be able to photograph a key and supply one or more corresponding images to hardware processor 110 (or any other suitable hardware processor) so that the hardware processor can process the image(s) to detect the key's bitting pattern and/or key blank type.” (Marsh, pg. 7, ¶[0090])
“In some embodiments, users can scan a key on their smartphone or tablet computer. Furthermore, in some embodiments, an application can be downloaded to assist users in capturing quality images of their key. From a mobile application or website platform, users can also request a duplicate of their scanned key be sent via mail to an address of their choosing, in some embodiments.” (Marsh, pg. 7, ¶[0091])
Thus, Marsh expressly teaches a key-identification application provided to the user on a mobile device, such as a smartphone or tablet computer, for capturing and supplying key images for processing to determine the key-blank type.
Regarding claim 3, Marsh in view of Chertok and further in view of Bass teach the computing system implemented method of Claim 1 wherein the one or more images of the cut key received from the user include:
“an image of a first surface of the cut key;” “and an image of a second surface of the cut key.” – Marsh teaches these limitations. Marsh teaches that a user may photograph a key and supply one or more corresponding images for processing to determine the key’s bitting pattern and key-blank type:
“a user can be able to photograph a key and supply one or more corresponding images to hardware processor 110 (or any other suitable hardware processor) so that the hardware processor can process the image(s) to detect the key's bitting pattern and/or key blank type.” (Marsh, pg. 7, ¶[0090])
Marsh further teaches capturing two images depicting different surfaces or views of the key:
“an imaging device 406 can be used to capture an image 502 of key 402. As another example, as shown in FIG. 6, an imaging device 408 can be used to capture an image 602 of key 402. Using these two images, parameters unique to each key can be detected in order to correctly determine both the bitting pattern and key blank type of the key.” (Marsh, pg. 6, ¶[0079])
Figure 5 depicts image 502 showing a side surface of the key, while Figure 6 depicts 602 showing a different surface or traverse view of the key. Marsh also expressly teaches detecting:
“a milling pattern of the key from one or more side views” (Marsh, pg. 2, ¶[0033])
Thus, Marsh teaches receiving a set of key images that include an image depicted a first surface of the cut key and an image depicting a second surface of the cut key. Marsh further teaches that obtaining the different views permits detection of parameters used to determine the key-blank type.
Regarding claim 4, Marsh in view of Chertok and further in view of Bass teach the computing system implemented method of Claim 1
“wherein the trained key blank identification machine learning model utilizes one or more image similarity algorithms to identify similarities between images of cut keys and images of key blanks, and further wherein the one or more similarity algorithms are selected from the group of image similarity algorithms consisting of: Siamese Neural Networks; and Convolutional Neural Networks.” – Marsh does not teach this limitation. Chertok, however, teaches this limitation. Chertok teaches a trained convolutional neural network that is applied to a query image and a reference image to determine the similarity between the respective images:
“Convolutional neural networks (CNN) are known in the art. These artificial networks of neurons can be trained by a training set of images and thereafter be employed for producing representations of an input image.” (Chertok, pg. 1, ¶[0002])
Chertok further states:
“The method includes the procedures of feeding a query image to a network, the network including a plurality of layers, and defining an output of each of the layers as a descriptor of the query image. The method also includes the procedures of feeding a reference image to the network and defining an output of each of the layers as a descriptor of the reference image and determining a descriptor similarity score for respective descriptors that were produced by the same layer of the network fed the query image and the reference image.” (Chertok, pg. 1, ¶[0005])
Chertok additionally teaches:
“Network executer 406 applies the network on the query image and on the reference image and records the output of each layer.” (Chertok, pg. 7, ¶[0069])
Chertok then states:
“Descriptor comparator 408 determines a descriptor similarity for each pair of respective descriptors.” (Chertok, pg. 7, ¶[0070])
Chertok further states:
“image comparator 412 defines the image similarity between the query image and the reference image as a function of the weighted descriptor similarities.” (Chertok, pg. 7, ¶[0071])
Finally, Chertok expressly identifies the neural-network type used for the disclosed image-similarity process:
“the methods and systems of the disclosed technique were exemplified by employing a CNN.” (Chertok, pg. 7, ¶[0073])
Thus, Chertok teaches a trained Convolutional Neural Network used as an image-similarity algorithm to identify similarities between a query image and a reference image. When applied to Marsh’s key-identification system, the trained CNN identifies similarities between images of cut keys and images of known key blanks. Because claim 4 permits “one or more” algorithm selected from the recited group, Chertok’s use of a Convolutional Neural Network satisfies the claimed selection.
A person of ordinary skill would have been motivated to implement Marsh’s machine-vision system using Chertok’s trained CNN image-similarity technique because Marsh expressly teaches using machine learning to improve machine-vision accuracy, while Chertok provides a known CNN-based method for comparing query and reference images. In the combination, the cut-key image would serve as the query image and stored key-blank images as reference images, predictably improving Marsh’s key-blank matching using Chertok’s technique according to its established function.
Regarding claim 5, Marsh in view of Chertok and further in view of Bass teach the computing system implemented method of Claim 1
“wherein the trained key blank identification machine learning model utilizes ” – Marsh teaches this limitation in part. Marsh teaches using image-derived characteristics of the key blade or bottom, including width variations and milling features, to determine the corresponding key-blank type. Marsh further teaches employing machine-learning algorithms to improve the accuracy of the machine-vision analysis:
“These width variations can be compared to known key blank data to determine a key blank type (or milling type). This data can be used to select the appropriate blank type during the key replication.” (Marsh, pg. 7, ¶[0081])
Marsh further states:
“Machine learning algorithms can also be employed to improve the accuracy and capabilities of the machine vision software.” (Marsh, pg. 7, ¶[0083])
Thus, Marsh teaches using machine-learning-enhanced image analysis of the key bottom or blade to determine a matching key-blank type.
Marsh does not teach these limitations and/or portions of:
“one or more weighting parameters to rank potential key blank matches, and further wherein similarities between portions of key images depicting the key bottom are weighted more heavily than similarities between portions of key images depicting the key head.”
Chertok, however, teaches these limitations and/or portions of:
“one or more weighting parameters to rank potential key blank matches, ” – Chertok teaches this limitation in part. Chertok teaches assigning respective weights to similarities between descriptors of a query image and a reference image and determining an overall image-similarity score as a function of the weighted similarities:
“a respective weight is assigned to each of the descriptor similarities.” (Chertok, pg. 6, ¶[0059])
Chertok further states:
“an image similarity between the query image and the reference image is defined as a function of weighted descriptor similarities” (Chertok, pg. 6, ¶[0060])
Chertok additionally teaches using the resulting image similarity for:
“various visual tasks such as image retrieval or machine vision” (Chertok, pg. 7, ¶[0071])
When Chertok’s technique is applied to Marsh’s submitted cut-key image and multiple candidate key-blank reference images, the resulting weighted image-similarity scores provides a basis for ranking the potential key-blank matches from most similar to least similar.
Regarding claim 6, Marsh in view of Chertok and further in view of Bass teach the computing system implemented method of Claim 5
“wherein similarities between portions of key images depicting the key bottom are weighted at 90% and similarities between portions of key images depicting the key head are weighted at 10%.” – Marsh teaches this limitation in part. As discussed regarding claim 5, Marsh teaches that width variations and milling characteristics located along the key blade or bottom are used to determine the corresponding key-blank type:
“These width variations can be compared to known key blank data to determine a key blank type (or milling type). This data can be used to select the appropriate blank type during the key replication.” (Marsh, pg. 7, ¶[0081])
Thus, Marsh teaches that characteristics of the key-bottom portion are pertinent to identifying the corresponding key blank.
Marsh does not teach these limitations and/or portions of:
“weighted at 90%”
“weighted at 10%.”
Chertok, however, teaches these limitations and/or portions of:
“weighted at ” and “weighted at .” – Chertok teaches this limitation in part. Chertok teaches assigning numerical weighting variables to respective image-descriptor similarities and determining those weights according to their effect on the known image-similarity result:
“
α
1
is the weight to be assigned (i.e., a variable) to the descriptor similarity between descriptors
D
1
i
and
D
1
j
(
S
1
).” (Chertok, pg. 4, ¶[0032])
Chertok further states:
“the weights
α
1
,
α
2
,
…
,
α
k
can be determined by regression, or by other methods or algorithms as known in the art.” (Chertok, pg. 4, ¶[0033])
Chertok additionally teaches that:
“a respective weight is assigned to each of the descriptor similarities.” (Chertok, pg. 6, ¶[0059])
Thus, Chertok teaches selecting and adjusting numerical weighting values.
Neither Marsh nor Chertok expressly teach these remaining limitations and/or portions of:
weighted at “90%” or “10%”
However, Chertok teaches assigning numerical weights to respective image-descriptor similarities and determining those weights through regression or other known algorithms. Marsh teaches that width variations and milling characteristics associated with the key-bottom or blade portion are sued to determine key-blank type. Thus, the prior art discloses both the general image-weighting conditions and the relative importance of the key-bottom information. The claimed allocation of 90% to key-bottom similarities and 10% to key-head similarities constitutes selection of particular proportions within Chertok’s known weighting scheme. It would have been obvious to determine workable or optimum weighting proportions through routine experimentation and to assign substantially greater weight to the key-bottom information identified by Marsh as pertinent to key-blank identification. See MPEP § 2144.05(II)(A).
A person or ordinary skill would have had reasonable expectation of success because Chertok expressly teaches that the numerical weights may be learned or determined according to their effect on image similarity, thereby allowing the weighting proportions to be adjusted and evaluated using known regression or training procedures. See MPEP § 2144.05(II)(B).
Regarding claim 7, Marsh in view of Chertok and further in view of Bass teach the computing system implemented method of Claim 1 wherein upon displaying the one or more images of the one or more matching key blanks and the corresponding matching key blank part numbers to the user:
“confirmation as to whether one of the matched key blank part numbers is the correct key blank part number is requested from the user; and confirmation that one of the matched key blank part numbers is the correct key blank part number is received from the user.” – Marsh does not these limitations. Bass, however, teaches these limitations. Bass teaches displaying a picture of the identified key blank specifically so that the user can verify wherever the identified blank is correct before proceeding:
“The visual display (i.e., the monitor, CRT or video system) would show a picture of the identified blank in order to permit visual verification by the user possibly by showing a 1 to 1 scale outline of the blank. Once confirmed, the user could also place the master key into an interface that translates the key cutting information to the key cutter” (Bass, pg. 7, ¶[0068])
Bass further teaches that the system displays:
“identification by manufacturer or part number (for example, ILCO number, ACE stock number, UPC code and/or the Jaw Selection)” (Marsh, pg. 5, ¶[0055])
Thus, Bass teaches displaying an identified key blank and corresponding part number for user verification and receiving confirmation before proceeding. In the combined system, Marsh’s input device would receive Bass’s user confirmation that one of the displayed matching key-blank part numbers is correct.
A person of ordinary skill would have been motivated to incorporate Bass’s visual-verification and confirmation operation into the Marsh-Chertok system so that the user verifies the identified key blank and part number before retrieval or cutting, predictably reducing incorrect blank selections and unsuccessful duplications.
Regarding claim 8, Marsh in view of Chertok and further in view of Bass teach the computing system implemented method of Claim 7
“wherein ” – Marsh teaches this limitation in part. Marsh teaches scanning a key to determine key-blank-type information and storing the resulting key information in a key template:
“This key scanning can detect the key bitting pattern and/or the key blank type in some embodiments.” (Marsh, pg. 5, ¶[0057])
Marsh further states:
“Finally, at 322, the information received at 316,318, and/or 320 can be stored. … For example, in some embodiments, this information can be stored in storage 108.” (Marsh, pg. 5, ¶[0063])
Marsh additionally states:
“that information, along with the bitting pattern and key blank type information, can be stored.” (Marsh, pg. 5, ¶[0064])
Marsh teaches that the stored key templates may include:
“descriptive names entered during a key template storage process, key images, etc.” (Marsh, pg. 6, ¶[0068])
Thus, Marsh teaches storing key images in a key-image storage system in association with identifying key-blank-type information, thereby teaching annotation of the stored images with key-identifying metadata.
Marsh does not teach these limitations and/or portions of:
“upon receiving confirmation that one of the matched key blank part numbers is the correct key blank part number … the correct key blank part number“
Bass, however, teaches these remaining limitations and/or portions of:
“upon receiving confirmation that one of the matched key blank part numbers is the correct key blank part number … the correct key blank part number“ – Bass teaches presenting an identified key blank for user confirmation, identifying the blank by part number, and updating database information based upon user input:
“The visual display (i.e., the monitor, CRT or video system) would show a picture of the identified blank in order to permit visual verification by the user, possibly by showing a 1 to 1 scale outline of the blank. Once confirmed, the user could also place the master key into an interface that translates the key cutting information to the key cutter” (Bass, pg. 7, ¶[0068])
Bass further teaches displaying:
“identification by manufacturer or part number (for example, ILCO number, ACE stock number, UPC code and/or the Jaw Selection)” (Bass, pg. 5, ¶[0055])
Bass additionally teaches:
“This database information can be updated periodically based upon use of the system, specific inputs by the user or according to a set schedule.” (Bass, pg. 3, ¶[0033])
Thus, Bass teaches receiving user confirmation of an identified key blank and corresponding part number and updating the database based upon that user input. In the combined system, Bas’s confirmed part number would be stored as identifying metadata associated with Marsh’s corresponding stored key image.
A person of ordinary skill would have been motivated to update Marsh’s stored key-image record with Bass’s user-confirmed part number because doing so would preserve the verified association between the cut-key image and the corresponding commercial key blank, facilitate later retrieval and duplication, and reduce repeated or erroneous key-blank identifications. The modification would constitute a predictable use of Bass’s user-driven database-update technique with Marsh’s existing key-template storage system.
Regarding claim 9, Marsh in view of Chertok and further in view of Bass teach the computing system implemented method of Claim 7
“” – Marsh teaches this limitation in part. Marsh expressly teaches providing the user with multiple mechanisms for obtaining the physical key blank required to duplicate the key:
“a user can determine a type of key blank required for replicating a particular key and obtain that type of key blank to use in replicating the key.” (Marsh, pg. 4, ¶[0053])
Marsh further states:
“the user can look up the type of key blank using a computer application, mobile application, or web platform.” (Marsh, pg. 4, ¶[0053])
Marsh additionally teaches that:
“a user can buy a blank of the required type at a retail location” (Marsh, pg. 4, ¶[0053])
“select a blank herself or acquire the appropriate blank from an attendant (e.g., a clerk).” (Marsh, pg. 4, ¶[0053])
Marsh also teaches an automated mechanism in which the processor controls the key-movement mechanism to:
“retrieve an appropriate key blank from a key repository and move the key to the key cutting and cleaning mechanism” (Marsh, pg. 5, ¶[0058])
Thus, Marsh teaches providing the user with purchasing, user-selection, attendant-assisted, and automated-retrieval mechanisms for obtaining a physical key blank corresponding to the identified key-blank type. Marsh does not expressly teach that these mechanisms are provided upon confirmation of a correct matched key-blank part number.
Marsh does no teach these limitations and/or portions of:
“wherein upon receiving confirmation that one of the matched key blank part numbers is the correct key blank part number, … to the correct key blank part number.” – Bass teaches displaying information identifying the selected key blank by:
“identification by manufacturer or part number (for example, ILCO number, ACE stock number, UPC code and/or the Jaw Selection)” (Marsh, pg. 5, ¶[0055])
Bass, however, teaches these remaining limitations and/or portions of:
“wherein upon receiving confirmation that one of the matched key blank part numbers is the correct key blank part number, … to the correct key blank part number.” – Bass teaches displaying information identifying the selected key blank by:
“manufacturer or part number (for example, ILCO number, ACE stock number, UPC code and/or the Jaw Selection)” (Marsh, pg. 5, ¶[0055])
Bass further teaches displaying a picture of the identified blank:
“in order to permit visual verification by the user” (Bass, pg. 7, ¶[0068])
and provides that:
“Once confirmed, the user could also place the master key into an interface that translates the key cutting information to the key cutter.” (Bass, pg. 7, ¶[0068])
Bass also teaches that:
“All identified key-blanks can be highlighted by the display rack for easy retrieval by the, user.” (Bass, pg. 7, ¶[0054])
Thus, Bass teaches confirming the identified key blank and corresponding part number before proceeding with retrieval or duplication. In the combined system, upon confirmation of Bass’s displayed key-blank image and part number, the user would be provided with Marsh’s mechanisms to purchase, select, acquire, or automatically retrieve the corresponding physical key blank.
A person of ordinary skill would have been motivated to provide Marsh’s blank-acquisition mechanisms after Bass’s confirmation step so that the user obtains the verified physical blank corresponding to the confirmed part number, predictably reducing incorrect blank selection and unsuccessful duplication.
Regarding claim 10, Marsh in view of Chertok and further in view of Bass teach the computing system implemented method of Claim 9 wherein upon providing the user with one or more mechanisms to obtain one or more physical key blanks corresponding to the correct key blank part number:
“at least one of the physical key blanks corresponding to the correct key blank part number is obtained by the user; and at least one of the obtained physical key blanks is cut by the user to produce a duplicate of the cut key.” – Marsh teaches this limitation. Marsh teaches that a user obtains the identified type of physical key blank for replication:
“a user can determine a type of key blank required for replicating a particular key and obtain that type of key blank to use in replicating the key.” (Marsh, pg. 4, ¶[0053])
Marsh further states that:
“a user can buy a blank of the required type at a retail location” (Marsh, pg. 4, ¶[0053])
or may:
“select a blank herself or acquire the appropriate blank from an attendant (e.g., a clerk).” (Marsh, pg. 4, ¶[0053])
Marsh further teaches a user-operated immediate-duplication process in which the corresponding physical key blank is retrieved and cut to produce the duplicate:
“the key can be replicated by the hardware processor 110: (a) controlling the key movement mechanism 114 to retrieve an appropriate key blank from a key repository and move the key to the key cutting and cleaning mechanism 116; and (b) controlling the key cutting and cleaning mechanism 116 to cut the key according to the detected bitting pattern and then clean the key to remove burrs, etc.” (Marsh, pg. 5, ¶[0058])
Marsh then provides the completed duplicate to the user:
“the process can cause the key to be dispensed to a user.” (Marsh, pg. 5, ¶[0059])
Thus, Marsh teaches that the user obtains the appropriate physical key blank and, through the user-operated duplication system, causes the obtained blank to be cut according to the cut key’s detected bitting pattern to produce a duplicate of the cut key.
Regarding claim 11, Marsh in view of Chertok and further in view of Bass, teach a computing system implemented method for identifying key blank part numbers comprising: aggregating a plurality of images of one or more cut keys; aggregating a plurality of images of one or more key blanks, wherein each of the key blanks correspond to at least one of the one or more cut keys; generating key blank identification training data based on the plurality of images of one or more cut keys and the plurality of images of one or more key blanks; generating a trained key blank identification machine learning model by training one or more machine learning models to identify key blanks using the key blank identification training data; receiving, by the trained key blank identification machine learning model, one or more images of a particular cut key from a user through a user interface of a key blank identification system; using the trained key blank identification machine learning model, identifying one or more key blank part numbers for one or more key blanks that match the particular cut key; based on the identified one or more key blank part numbers, retrieving one or more images of the one or more matching key blanks from a key image database; displaying the one or more images of the one or more matching key blanks and the corresponding key blank part numbers to the user through a user interface of the key blank identification system; receiving user feedback from the user through the user interface of the key blank identification system; upon user confirmation of identification of a correct key blank, annotating the one or more images of the particular cut key with the key blank part number corresponding to the correct key blank; incorporating the user feedback into the key blank identification training data to generate a refined trained key blank identification machine learning model; and utilizing the refined key blank identification machine learning model to identify unique key blank part numbers that correspond with a plurality of cut keys.
Regarding claims 12-20
Claims 12-19 depend, directly or indirectly, from independent method claim 11 and further limit the training, image-acquisition, image-similarity, weighting, user-feedback, annotation, and model-refinement operations recited therein. Claim 20 is an independent system claim that recites system components configured to perform substantially the same training, identification, feedback, annotation, and model-refinement operations recited in claim 11, together with additional limitations corresponding to claims 12-14 and 18-19.
Marsh, Chertok, and Bass teach the underlying image-acquisition, key-blank identification, machine-learning, image-similarity, key-image database, user-confirmation, annotation, and model-refinement operations for the reasons discussed above with respect to claim 11. The references implement those operations using processor-based computing systems including processors, storage, databases, machine-learning models, user interfaces, and applications.
Claim 12 further requires that generating the key-blank-identification training data includes annotating one or more cut-key images with a unique key-blank part number. Chertok teaches supervised image training using labeled images, Marsh teaches storing cut-key images with associated key-blank information, and Bass teaches identifying key blanks using unique manufacturer, stock, or part numbers. It would have been obvious to use Bass’s key-blank part number as the supervised label associated with Marsh’s corresponding cut-key image when generating Chertok’s training data. Accordingly, claim 12 is obvious over Marsh in view of Chertok and further in view of Bass.
Claim 13 recites first- and second-surface images of the particular cut key. Marsh teaches capturing images depicting different surfaces or views of a key for determining the key-blank type for the reasons discussed above with respect to claim 3. Accordingly, claim 13 is obvious over Marsh in view of Chertok and further in view of Bass.
Claim 14 recites an image-similarity algorithm selected from Siamese Neural Networks and Convolutional Neural Networks. Chertok teaches employing a trained convolutional neural network to compare query and reference images for the reasons discussed above with respect to claim 4. Accordingly, claim 14 is obvious over Marsh in view of Chertok and further in view of Bass.
Claim 15 recites using one or more weighting parameters to rank potential key-blank matches. Chertok teaches assigning learned weighting parameters to image-descriptor similarities and determining overall image similarity from the weighted similarities for the reasons discussed above with respect to claim 5. Applying the resulting similarity values to Marsh’s candidate key blanks would rank the candidates according to similarity. Accordingly, claim 15 is obvious over Marsh in view of Chertok and further in view of Bass.
Claim 16 recites weighting similarities depicting the key bottom more heavily than similarities depicting the key head. Marsh teaches relying on blade-width and milling characteristics to determine key-blank type, while Chertok teaches assigning different weights to respective image similarities, for the reasons discussed above with respect to claim 5. Accordingly, claim 16 is obvious over Marsh in view of Chertok and further in view of Bass.
Claim 17 recites weighting the key-bottom and key-head similarities at 90% and 10%, respectively. Chertok teaches learned numerical weighting parameters, and selection of the claimed percentages would have been an obvious optimization of a result-effective variable for the reasons discussed above with respect to claim 6. Accordingly, claim 17 is obvious over Marsh in view of Chertok and further in view of Bass.
Claim 18 recites feedback indicating whether one of the matching key-blank part numbers is correct. Marsh teaches receiving a user’s selection from displayed key choices, while Bass teaches visual verification of an identified key blank and its corresponding part number, for the reasons discussed above with respect to claim 7. Accordingly, claim 18 is obvious over Marsh in view of Chertok and further in view of Bass.
Claim 19 recites, upon the user’s indication of a correct key-blank part number, annotating the images of the particular cut key with the correct part number and storing the annotated images in the key-image database. Marsh teaches storing key images with associated key-blank information, while Bass teaches user confirmation and part-number identification, for the reasons discussed above with respect to claim 8. It would have been obvious to store Bass’s confirmed part number with Marsh’s corresponding cut-key images to preserve the verified association for subsequent retrieval and model refinement. Accordingly, claim 19 is obvious over Marsh in view of Chertok and further in view of Bass.
Claim 20 recites a system comprising a key-image database, one or more machine-learning models, a key-blank-identification application, at least one processor, and at least one memory storing instructions that cause the processor to perform substantially the same operations recited in claim 11 and the additional operations recited in claims 12-14 and 18-19.
Marsh teaches a processor, storage, user interfaces and applications, receipt and storage of key images, and machine-learning techniques to determine key type and bitting pattern. Chertok teaches a trained convolutional neural network, data storage containing query and reference images, and processor-executed image-similarity operations. Bass teaches a key-blank database containing key-blank pictures and corresponding part numbers, user verification, and adaptation based on mistakes or user inputs.
It would have been obvious to store instructions for performing the combined Marsh-Chertok-Bass operations in memory and execute those instructions using a processor because the disclosed identification, image-comparison, database, feedback, and model-training operations are computer-implemented. Expressing the previously established method operations as processor-configured system functions does not, without more, impart a patentable distinction over the prior-art combination. Accordingly, claim 20 is obvious over Marsh in view of Chertok and further in view of Bass.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Paul Coleman whose telephone number is (571)272-4687. The examiner can normally be reached Mon-Fri.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, David Yi can be reached at (571) 270-7519. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/PAUL COLEMAN/ Examiner, Art Unit 2126
/DAVID YI/ Supervisory Patent Examiner, Art Unit 2126
Prosecution Insights