DETAILED ACTION
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 .
Response to Amendments
Applicant’s remarks, see page 8, filed 06/25/2026, with respect to drawing objections that did not include reference sign(s) mentioned in the description, submitted in the non-final office action dated 04/01/2026, have been fully considered and are persuasive due to amendments to the specification to refer to the correct reference signs for method 200, and steps 210, 220 and 230 in compliance with Fig. 2A. Thus, drawing objections regarding the reference signs for method 200, and steps 210, 220 and 230 in compliance with Fig. 2A have been withdrawn.
Applicant’s remarks, see pages 7-9, filed 06/26/2026, with respect to claim limitations interpreted under 35 U.S.C. 112(f) in claims 1 and 11, along with rejections under 35 U.S.C. 112(a) and 35 U.S.C. 112(b), submitted in the non-final office action dated 04/01/2026, have been fully considered and are persuasive in light of claim amendments. Thus, the interpretation of “classification unit” under 35 U.S.C. 112(f) and rejections under 35 U.S.C. 112(a) and 35 U.S.C. 112(b) in claims 1 and 11 have been withdrawn.
Applicant’s remarks, see pages 9-10, filed 06/26/2026, with respect to the rejection of claims 1-20 under 35 U.S.C. 101, have been fully considered and are persuasive in light of claim amendments. Thus, the rejection of claims 1-20 under 35 U.S.C. 101 has been withdrawn.
Response to Arguments
Applicant’s arguments, see remarks, filed 06/26/2026, with respect to claims 1-20, have been fully considered, but are moot because the arguments do not apply to the current references and current combinations of references being used in the current rejection.
Drawing Objections
The drawings are objected to as failing to comply with 37 CFR 1.84(p)(5) because they do not include the following reference sign(s) mentioned in the description:
Paragraph [0083] states “Figure 3B is an example of a method 250…”should read “Figure 2B is an example of a method 250…” in order to refer to the right figure as the drawings do not include a Figure 3B.
Corrected drawing sheets in compliance with 37 CFR 1.121(d) are required in reply to the Office action to avoid abandonment of the application. Any amended replacement drawing sheet should include all of the figures appearing on the immediate prior version of the sheet, even if only one figure is being amended. Each drawing sheet submitted after the filing date of an application must be labeled in the top margin as either “Replacement Sheet” or “New Sheet” pursuant to 37 CFR 1.121(d). If the changes are not accepted by the examiner, the applicant will be notified and informed of any required corrective action in the next Office action. The objection to the drawings will not be held in abeyance.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
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 of this title, 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 1-4, 6-14, and 16-20 are rejected under 35 U.S.C. 103 as being unpatentable over --ZHAO (WO 2024197443 A1), hereinafter referenced as ZHAO in view of KIM (US 20220067405 A1), hereinafter referenced as KIM.
Regarding claim 1, ZHAO teaches a method of dynamic classification for autonomous driving (Fig. 8, Paragraph [0084] – ZHAO discloses FIGURE 8 is a flow diagram illustrating a processor-implemented method 800 for dynamic class-incremental learning without forgetting. Paragraph [0088] – ZHAO further discloses the ANN model may be included at a user device. The user device may comprise a mobile computing device such as a smartphone, an autonomous vehicle or other mobile computing device, for instance.), the method comprises:
producing, by a classification neural network associated with a driving of a vehicle (Fig. 5, Paragraph [0065] – ZHAO discloses the ANN model may, in turn, be distributed to one or more user devices 520a, 520b. The user device ( e.g., 520a or 520b) may be a mobile electronic device ( e.g., a smartphone, a wearable computing device ( e.g., smartglasses)) or an autonomous driving vehicle, for example.),
a catalog sample representative vector representing a catalog sample of a new class for autonomous driving being unidentified by the classification neural network (Fig. 8, Paragraph [0087] – ZHAO discloses the processor generates by the first ANN, a new class without retraining the first ANN. The ANN may generate an embedding (e.g., using the backbone 504 and projector 506) for representative samples of a new class.)
at the time of the driving (Fig. 5, Paragraph [0065] – ZHAO discloses the ANN model may, in turn, be distributed to one or more user devices 520a, 520b. The user device ( e.g., 520a or 520b) may be a mobile electronic device ( e.g., a smartphone, a wearable computing device ( e.g., smartglasses)) or an autonomous driving vehicle, for example.),
generating, by the classification neural network, a new image-based representative vector (Fig. 8, Paragraph [0085] – ZHAO discloses at block 804, the processor extracts, by the first ANN, features of the input to generate a representation of the input. For example, as discussed with respect to FIGURE 5, the backbone 504 may process the input 502, extracting features of the input 502 to generate a representation of the input 502.)
representing an image of an entity of the new class (Fig. 8, Paragraph [0084] – ZHAO discloses at block 802, the processor receives, by a first artificial neural network (ANN), an input. The input may comprise image data, video data, sensor data, sequence data, or other type of data.);
Although ZHAO further teaches and classifying, by the classification neural network, the entity as being associated with the new class (Fig. 5, Paragraph [0067] – ZHAO discloses the on-device ANN may generate an embedding (e.g., using the backbone 526 and projector 528) for representative samples of a new class. The embedding may then be saved and added to the index ( e.g., 522a or 522b ). Classifications may be made by a nearest-mean-of-samples rule.).
ZHAO fails to explicitly teach wherein the catalog sample is a diagram or graphical illustration of an item that differs from an actual image of the item; wherein the classification neural network is trained by a training process that comprises clustering image-based representative vectors of a given class based on distances between the image-based representative vectors of the given class and a catalog sample representative vector representing a catalog sample of the given class;
However, KIM explicitly teaches wherein the catalog sample is a diagram or graphical illustration of an item that differs from an actual image of the item (Fig. 2, Paragraph [0045] – KIM discloses the road sign template may refer to a prototype image that may be extracted from a convention document (e.g., MUTCD) [wherein the road sign template/prototype image is the catalog sample]. The road sign may refer to a real physical sign on the image collected by a camera [wherein the road sign is an actual image of the item].);
wherein the classification neural network is trained by a training process that comprises clustering image-based representative vectors of a given class based on distances between the image-based representative vectors of the given class and a catalog sample representative vector representing a catalog sample of the given class (Fig. 5, Paragraph [0020] – KIM discloses embodiments below described a system that utilizes knowledge graph and machine learning models that classify various road signs and/or classify visual attributes of road signs, which can calculate distances between a real sign on an image and sign prototypes represented in a latent space to provide ranked sign prototypes matching with a real sign on an image—to assist annotators in classifying road signs effectively. Paragraph [0058] – KIM further discloses to improve the one-shot classifier, we introduce metric learning, such as triplet loss, during training to further separate different classes in the embedding space.);
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date the claimed invention was made to combine the teachings of ZHAO of having a method of dynamic classification for autonomous driving, the method comprises: producing, by a classification neural network associated with a driving of a vehicle, a catalog sample representative vector representing a catalog sample of a new class for autonomous driving being unidentified by the classification neural network at the time of the driving, generating, by the classification neural network, a new image-based representative vector representing an image of an entity of the new class; and classifying, by the classification neural network, the entity as being associated with the new class, with the teachings of KIM of having wherein the catalog sample is a diagram or graphical illustration of an item that differs from an actual image of the item; wherein the classification neural network is trained by a training process that comprises clustering image-based representative vectors of a given class based on distances between the image-based representative vectors of the given class and a catalog sample representative vector representing a catalog sample of the given class;
Wherein having ZHAO’s method of dynamic classification for autonomous driving wherein the catalog sample is a diagram or graphical illustration of an item that differs from an actual image of the item; wherein the classification neural network is trained by a training process that comprises clustering image-based representative vectors of a given class based on distances between the image-based representative vectors of the given class and a catalog sample representative vector representing a catalog sample of the given class;
The motivation behind the modification would have been to obtain an improved method of identifying new classes that enables efficient resource usage and reduces manual annotation efforts, since both ZHAO and KIM relate to image processing and classification using machine learning models, wherein ZHAO relates to dynamic class incremental learning without forgetting that may beneficially combine embedding representation learning plus index and search for incremental learning without forgetting on-device, as such, enabling efficient resource usage by reducing, and in some aspects, eliminating retraining from scratch at the arrival of new data, and KIM relates to machine learning and road sign recognition wherein utilizing a knowledge graph approach can reduce sign search space; the system can propose the correct single candidate by and large, reducing the human search effort fairly. Please see ZHAO (WO 2024197443 A1), Paragraph [0029, 0079], KIM (US 20220067405 A1), Paragraph [0020].
Regarding claim 2, ZHAO in view of KIM teach the method according to claim 1,
ZHAO further teaches comprising dynamically updating a set of catalog representative vectors (Fig. 8, Paragraph [0080] – ZHAO discloses the embeddings for representative items of new classes may be computed and added to an index. Accordingly, an unlimited new number of classes may be added to the index without having to retrain the on-device ANN model.),
and absent weights amendments of the classification neural network (Fig. 8, Paragraph [0087] – ZHAO discloses the processor generates by the first ANN, a new class without retraining the first ANN.).
Regarding claim 3, ZHAO in view of KIM teach the method according to claim 2,
ZHAO further teaches comprising dynamically updating a set of catalog representative vectors (Fig. 8, Paragraph [0080] – ZHAO discloses the embeddings for representative items of new classes may be computed and added to an index. Accordingly, an unlimited new number of classes may be added to the index without having to retrain the on-device ANN model.)
associated with a scenario being faced by the vehicle (Fig. 8, Paragraph [0084] – ZHAO discloses the processor receives, by a first artificial neural network (ANN), an input. The input may comprise image data, video data, sensor data, sequence data, or other type of data. The input may be supplied directly from a user device [wherein user device is the vehicle, see also Paragraph [0065]] or may be streamed via the Internet, for example.).
Regarding claim 4, ZHAO in view of KIM teach the method according to claim 1,
ZHAO fails to explicitly teach wherein an obtained representative vector of an existing class is classified with a confidence level, wherein the confidence level is calculated based on a distance between the obtained representative vector to other obtained representative vectors.
However, KIM explicitly teaches wherein an obtained representative vector of an existing class is classified with a confidence level (Fig. 5, Paragraph [0055] – KIM discloses the system may integrate a machine learning model such as one-shot classifier to predict top-K road sign template candidates that are similar to a cropped image patch containing a road sign. The inputs for this model may be (1) a cropped image patch around the bounding box that the annotator draws on the real road image, and (2) sign templates filtered by the Road Sign Knowledge Graph. These two inputs above may be encoded into the latent space, and the nearest neighbor classification may be used to rank the road sign templates. See also paragraph [0020, 0058].),
wherein the confidence level is calculated based on a distance between the obtained representative vector to other obtained representative vectors (Fig. 5, Paragraph [0055] – KIM discloses the system may integrate a machine learning model such as one-shot classifier to predict top-K road sign template candidates that are similar to a cropped image patch containing a road sign. The inputs for this model may be (1) a cropped image patch around the bounding box that the annotator draws on the real road image, and (2) sign templates filtered by the Road Sign Knowledge Graph. These two inputs above may be encoded into the latent space, and the nearest neighbor classification may be used to rank the road sign templates. See also paragraph [0020, 0058].).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date the claimed invention was made to combine the teachings of ZHAO in view of KIM of having a method of dynamic classification for autonomous driving, the method comprises: producing, by a classification neural network associated with a driving of a vehicle, a catalog sample representative vector representing a catalog sample of a new class for autonomous driving being unidentified by the classification neural network at the time of the driving, generating, by the classification neural network, a new image-based representative vector representing an image of an entity of the new class; and classifying, by the classification neural network, the entity as being associated with the new class, with the teachings of KIM of having wherein an obtained representative vector of an existing class is classified with a confidence level, wherein the confidence level is calculated based on a distance between the obtained representative vector to other obtained representative vectors.
Wherein having ZHAO’s method of dynamic classification for autonomous driving wherein an obtained representative vector of an existing class is classified with a confidence level, wherein the confidence level is calculated based on a distance between the obtained representative vector to other obtained representative vectors.
The motivation behind the modification would have been to obtain an improved method of identifying new classes that enables efficient resource usage and reduces manual annotation efforts, since both ZHAO and KIM relate to image processing and classification using machine learning models, wherein ZHAO relates to dynamic class incremental learning without forgetting that may beneficially combine embedding representation learning plus index and search for incremental learning without forgetting on-device, as such, enabling efficient resource usage by reducing, and in some aspects, eliminating retraining from scratch at the arrival of new data, and KIM relates to machine learning and road sign recognition wherein utilizing a knowledge graph approach can reduce sign search space; the system can propose the correct single candidate by and large, reducing the human search effort fairly. Please see ZHAO (WO 2024197443 A1), Paragraph [0029, 0079], KIM (US 20220067405 A1), Paragraph [0020].
Regarding claim 6, ZHAO in view of KIM teach the method according to claim 1,
ZHAO further teaches wherein the training process comprises applying, for the given class, a loss process that is operative to reduce an angle between image-based representative vectors of the given class and a catalog sample representing vector of the given class (Fig. 5, Paragraph [0063] – ZHAO discloses input 502 may, for instance, include pairs of images or a batch of samples. The backbone 504 may process the input 502, extracting features of the input 502 to generate a representation of the input 502. Paragraph [0065] – ZHAO discloses the input 502 may be mined for pairs of samples (may also be referred to as "examples") and weighted for pair-based loss functions. The example architecture 500 may learn to minimize the distance (e.g., cosine distance [wherein cosine distance is a loss process that is operative to reduce an angle]) between similar samples and to maximize the distance between dissimilar samples.)
and increase the angle between the image-based representative vectors of the given class and other catalog sample representative vectors of other classes (Fig. 5, Paragraph [0063] – ZHAO discloses input 502 may, for instance, include pairs of images or a batch of samples. The backbone 504 may process the input 502, extracting features of the input 502 to generate a representation of the input 502. Paragraph [0065] – ZHAO discloses the input 502 may be mined for pairs of samples (may also be referred to as "examples") and weighted for pair-based loss functions. The example architecture 500 may learn to minimize the distance (e.g., cosine distance [wherein cosine distance is a loss process that is operative to increase the angle]) between similar samples and to maximize the distance between dissimilar samples.).
Regarding claim 7, ZHAO in view of KIM teach the method according to claim 1,
ZHAO fails to explicitly teach wherein the training process involves applying a triple loss process.
However, KIM explicitly teaches wherein the training process involves applying a triple loss process (Fig. 5, Paragraph [0058] – KIM discloses to improve the one-shot classifier, we introduce metric learning, such as triplet loss, during training to further separate different classes in the embedding space.).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date the claimed invention was made to combine the teachings of ZHAO in view of KIM of having a method of dynamic classification for autonomous driving, the method comprises: producing, by a classification neural network associated with a driving of a vehicle, a catalog sample representative vector representing a catalog sample of a new class for autonomous driving being unidentified by the classification neural network at the time of the driving, wherein the classification neural network is trained by a training process that comprises clustering image-based representative vectors of a given class based on distances between the image-based representative vectors of the given class and a catalog sample representative vector representing a catalog sample of the given class; with the teachings of KIM of having wherein the training process involves applying a triple loss process.
Wherein having ZHAO’s method of dynamic classification for autonomous driving wherein the training process involves applying a triple loss process.
The motivation behind the modification would have been to obtain an improved method of identifying new classes that enables efficient resource usage and reduces manual annotation efforts, since both ZHAO and KIM relate to image processing and classification using machine learning models, wherein ZHAO relates to dynamic class incremental learning without forgetting that may beneficially combine embedding representation learning plus index and search for incremental learning without forgetting on-device, as such, enabling efficient resource usage by reducing, and in some aspects, eliminating retraining from scratch at the arrival of new data, and KIM relates to machine learning and road sign recognition wherein utilizing a knowledge graph approach can reduce sign search space; the system can propose the correct single candidate by and large, reducing the human search effort fairly. Please see ZHAO (WO 2024197443 A1), Paragraph [0029, 0079], KIM (US 20220067405 A1), Paragraph [0020].
Regarding claim 8, ZHAO in view of KIM teach the method according to claim 1,
ZHAO fails to explicitly teach wherein the producing comprises performing a one-shot learning process.
However, KIM explicitly teaches wherein the producing comprises performing a one-shot learning process (Fig. 5, Paragraph [0058] – KIM discloses next classifier 515 may be a one-shot classifier based on an encoder, such as a vibrational prototyping encoder, which learns a similarity metric to classify images at test time using a single prototype for each road sign within the candidate classes.).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date the claimed invention was made to combine the teachings of ZHAO in view of KIM of having a method of dynamic classification for autonomous driving, the method comprises: producing, by a classification neural network associated with a driving of a vehicle, a catalog sample representative vector representing a catalog sample of a new class for autonomous driving being unidentified by the classification neural network at the time of the driving, with the teachings of KIM of having wherein the producing comprises performing a one-shot learning process.
Wherein having ZHAO’s method of dynamic classification for autonomous driving wherein the producing comprises performing a one-shot learning process.
The motivation behind the modification would have been to obtain an improved method of identifying new classes that enables efficient resource usage and reduces manual annotation efforts, since both ZHAO and KIM relate to image processing and classification using machine learning models, wherein ZHAO relates to dynamic class incremental learning without forgetting that may beneficially combine embedding representation learning plus index and search for incremental learning without forgetting on-device, as such, enabling efficient resource usage by reducing, and in some aspects, eliminating retraining from scratch at the arrival of new data, and KIM relates to machine learning and road sign recognition wherein utilizing a knowledge graph approach can reduce sign search space; the system can propose the correct single candidate by and large, reducing the human search effort fairly. Please see ZHAO (WO 2024197443 A1), Paragraph [0029, 0079], KIM (US 20220067405 A1), Paragraph [0020].
Regarding claim 9, ZHAO in view of KIM teach the method according to claim 1,
ZHAO fails to explicitly teach wherein the producing comprises performing a few-shot learning process.
However, KIM explicitly teaches wherein the producing comprises performing a few-shot learning process (Fig. 6C, Paragraph [0062] – KIM discloses The system may output road sign icons along with the image patch (cropped based on human annotator's input of a geometric shape) that are used for few shot learning classifier that produces top K candidates. The search space (# of candidates) may be reduced by the Knowledge Graph provides a better prediction in few shot learning classifier.).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date the claimed invention was made to combine the teachings of ZHAO in view of KIM of having a method of dynamic classification for autonomous driving, the method comprises: producing, by a classification neural network associated with a driving of a vehicle, a catalog sample representative vector representing a catalog sample of a new class for autonomous driving being unidentified by the classification neural network at the time of the driving, with the teachings of KIM of having wherein the producing comprises performing a few-shot learning process.
Wherein having ZHAO’s method of dynamic classification for autonomous driving wherein the producing comprises performing a few-shot learning process.
The motivation behind the modification would have been to obtain an improved method of identifying new classes that enables efficient resource usage and reduces manual annotation efforts, since both ZHAO and KIM relate to image processing and classification using machine learning models, wherein ZHAO relates to dynamic class incremental learning without forgetting that may beneficially combine embedding representation learning plus index and search for incremental learning without forgetting on-device, as such, enabling efficient resource usage by reducing, and in some aspects, eliminating retraining from scratch at the arrival of new data, and KIM relates to machine learning and road sign recognition wherein utilizing a knowledge graph approach can reduce sign search space; the system can propose the correct single candidate by and large, reducing the human search effort fairly. Please see ZHAO (WO 2024197443 A1), Paragraph [0029, 0079], KIM (US 20220067405 A1), Paragraph [0020].
Regarding claim 10, ZHAO in view of KIM teach the method according to claim 1,
Although ZHAO further teaches wherein the image of the entity of the new class (Fig. 8, Paragraph [0085]),
ZHAO fails to explicitly teach wherein the image of the entity of the new class is a cropped image of the entity of the new class.
However, KIM explicitly teaches wherein the image of the entity of the new class is a cropped image of the entity of the new class (Fig. 5, Paragraph [0055] – KIM discloses the system may integrate a machine learning model such as one-shot classifier to predict top-K road sign template candidates that are similar to a cropped image patch containing a road sign. The inputs for this model may be (1) a cropped image patch around the bounding box that the annotator draws on the real road image, and (2) sign templates filtered by the Road Sign Knowledge Graph. These two inputs above may be encoded into the latent space, and the nearest neighbor classification may be used to rank the road sign templates.).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date the claimed invention was made to combine the teachings of ZHAO in view of KIM of having a method of dynamic classification for autonomous driving, the method comprises: generating, by the classification neural network, a new image-based representative vector representing an image of an entity of the new class; with the teachings of KIM of having wherein the image of the entity of the new class is a cropped image of the entity of the new class.
Wherein having ZHAO’s method of dynamic classification for autonomous driving wherein the image of the entity of the new class is a cropped image of the entity of the new class.
The motivation behind the modification would have been to obtain an improved method of identifying new classes that enables efficient resource usage and reduces manual annotation efforts, since both ZHAO and KIM relate to image processing and classification using machine learning models, wherein ZHAO relates to dynamic class incremental learning without forgetting that may beneficially combine embedding representation learning plus index and search for incremental learning without forgetting on-device, as such, enabling efficient resource usage by reducing, and in some aspects, eliminating retraining from scratch at the arrival of new data, and KIM relates to machine learning and road sign recognition wherein utilizing a knowledge graph approach can reduce sign search space; the system can propose the correct single candidate by and large, reducing the human search effort fairly. Please see ZHAO (WO 2024197443 A1), Paragraph [0029, 0079], KIM (US 20220067405 A1), Paragraph [0020].
Regarding claim 11, ZHAO teaches a non-transitory computer readable medium for dynamic classification (Fig. 1, Paragraph [0107] – ZHAO discloses the functions may be stored or transmitted over as one or more instructions or code on a computer-readable medium. Thus, in some aspects, computer-readable media may comprise non-transitory computer-readable media (e.g. tangible media). See also Paragraph [0030].)
for autonomous driving (Fig. 8, Paragraph [0088] – ZHAO further discloses the ANN model may be included at a user device. The user device may comprise a mobile computing device such as a smartphone, an autonomous vehicle or other mobile computing device, for instance.),
the non-transitory computer readable medium stores instructions (Fig. 1, Paragraph [0107]) for:
producing, by a classification neural network associated with a driving of a vehicle, a catalog sample representative vector representing a catalog sample of a new class for autonomous driving being unidentified by the classification neural network (Fig. 8, Paragraph [0087] – ZHAO discloses the processor generates by the first ANN, a new class without retraining the first ANN. The ANN may generate an embedding (e.g., using the backbone 504 and projector 506) for representative samples of a new class.)
at the time of the driving (Fig. 5, Paragraph [0065] – ZHAO discloses the ANN model may, in turn, be distributed to one or more user devices 520a, 520b. The user device (e.g., 520a or 520b) may be a mobile electronic device (e.g., a smartphone, a wearable computing device (e.g., smartglasses)) or an autonomous driving vehicle, for example.),
generating, by the classification neural network, a new image-based representative vector (Fig. 8, Paragraph [0085] – ZHAO discloses at block 804, the processor extracts, by the first ANN, features of the input to generate a representation of the input. For example, as discussed with respect to FIGURE 5, the backbone 504 may process the input 502, extracting features of the input 502 to generate a representation of the input 502.)
representing an image of an entity of the new class (Fig. 8, Paragraph [0084] – ZHAO discloses at block 802, the processor receives, by a first artificial neural network (ANN), an input. The input may comprise image data, video data, sensor data, sequence data, or other type of data.);
Although ZHAO further teaches and classifying, by the classification neural network, the entity as being associated with the new class (Fig. 5, Paragraph [0067] – ZHAO discloses the on-device ANN may generate an embedding (e.g., using the backbone 526 and projector 528) for representative samples of a new class. The embedding may then be saved and added to the index ( e.g., 522a or 522b ). Classifications may be made by a nearest-mean-of-samples rule.).
ZHAO fails to explicitly teach wherein the catalog sample is a diagram or graphical illustration of an item that differs from an actual image of the item; wherein the classification neural network is trained by a training process that comprises clustering image-based representative vectors of a given class based on distances between the image-based representative vectors of the given class and a catalog sample representative vector representing a catalog sample of the given class;
However, KIM explicitly teaches wherein the catalog sample is a diagram or graphical illustration of an item that differs from an actual image of the item (Fig. 2, Paragraph [0045] – KIM discloses the road sign template may refer to a prototype image that may be extracted from a convention document (e.g., MUTCD) [wherein the road sign template/prototype image is the catalog sample]. The road sign may refer to a real physical sign on the image collected by a camera [wherein the road sign is an actual image of the item].);
wherein the classification neural network is trained by a training process that comprises clustering image-based representative vectors of a given class based on distances between the image-based representative vectors of the given class and a catalog sample representative vector representing a catalog sample of the given class (Fig. 5, Paragraph [0020] – KIM discloses embodiments below described a system that utilizes knowledge graph and machine learning models that classify various road signs and/or classify visual attributes of road signs, which can calculate distances between a real sign on an image and sign prototypes represented in a latent space to provide ranked sign prototypes matching with a real sign on an image—to assist annotators in classifying road signs effectively. Paragraph [0058] – KIM further discloses to improve the one-shot classifier, we introduce metric learning, such as triplet loss, during training to further separate different classes in the embedding space.);
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date the claimed invention was made to combine the teachings of ZHAO of having a non-transitory computer readable medium for dynamic classification for autonomous driving, the non-transitory computer readable medium stores instructions for: producing, by a classification neural network associated with a driving of a vehicle, a catalog sample representative vector representing a catalog sample of a new class for autonomous driving being unidentified by the classification neural network at the time of the driving, generating, by the classification neural network, a new image-based representative vector representing an image of an entity of the new class; and classifying, by the classification neural network, the entity as being associated with the new class, with the teachings of KIM of having wherein the catalog sample is a diagram or graphical illustration of an item that differs from an actual image of the item; wherein the classification neural network is trained by a training process that comprises clustering image-based representative vectors of a given class based on distances between the image-based representative vectors of the given class and a catalog sample representative vector representing a catalog sample of the given class;
Wherein having ZHAO’s non-transitory computer readable medium for dynamic classification for autonomous driving of dynamic classification for autonomous driving wherein the catalog sample is a diagram or graphical illustration of an item that differs from an actual image of the item; wherein the classification neural network is trained by a training process that comprises clustering image-based representative vectors of a given class based on distances between the image-based representative vectors of the given class and a catalog sample representative vector representing a catalog sample of the given class;
The motivation behind the modification would have been to obtain an improved method of identifying new classes that enables efficient resource usage and reduces manual annotation efforts, since both ZHAO and KIM relate to image processing and classification using machine learning models, wherein ZHAO relates to dynamic class incremental learning without forgetting that may beneficially combine embedding representation learning plus index and search for incremental learning without forgetting on-device, as such, enabling efficient resource usage by reducing, and in some aspects, eliminating retraining from scratch at the arrival of new data, and KIM relates to machine learning and road sign recognition wherein utilizing a knowledge graph approach can reduce sign search space; the system can propose the correct single candidate by and large, reducing the human search effort fairly. Please see ZHAO (WO 2024197443 A1), Paragraph [0029, 0079], KIM (US 20220067405 A1), Paragraph [0020].
Regarding claim 12, ZHAO in view of KIM teach the non-transitory computer readable medium according to claim 11,
ZHAO further teaches that stores instructions (Fig. 1, Paragraph [0107]) for dynamically updating a set of catalog representative vectors (Fig. 8, Paragraph [0080] – ZHAO discloses the embeddings for representative items of new classes may be computed and added to an index. Accordingly, an unlimited new number of classes may be added to the index without having to retrain the on-device ANN model.),
and absent weights amendments of the classification neural network (Fig. 8, Paragraph [0087] – ZHAO discloses the processor generates by the first ANN, a new class without retraining the first ANN.).
Regarding claim 13, ZHAO in view of KIM teach the non-transitory computer readable medium according to claim 12,
ZHAO further teaches that stores instructions (Fig. 1, Paragraph [0107]) for dynamically updating a set of catalog representative vectors (Fig. 8, Paragraph [0080] – ZHAO discloses the embeddings for representative items of new classes may be computed and added to an index. Accordingly, an unlimited new number of classes may be added to the index without having to retrain the on-device ANN model.)
associated with a scenario being faced by the vehicle (Fig. 8, Paragraph [0084] – ZHAO discloses the processor receives, by a first artificial neural network (ANN), an input. The input may comprise image data, video data, sensor data, sequence data, or other type of data. The input may be supplied directly from a user device [wherein user device is the vehicle, see also Paragraph [0065]] or may be streamed via the Internet, for example.).
Regarding claim 14, ZHAO in view of KIM teach the non-transitory computer readable medium according to claim 11,
ZHAO fails to explicitly teach wherein an obtained representative vector of an existing class is classified with a confidence level, wherein the confidence level is calculated based on a distance between the obtained representative vector to other obtained representative vectors.
However, KIM explicitly teaches wherein an obtained representative vector of an existing class is classified with a confidence level (Fig. 5, Paragraph [0055] – KIM discloses the system may integrate a machine learning model such as one-shot classifier to predict top-K road sign template candidates that are similar to a cropped image patch containing a road sign. The inputs for this model may be (1) a cropped image patch around the bounding box that the annotator draws on the real road image, and (2) sign templates filtered by the Road Sign Knowledge Graph. These two inputs above may be encoded into the latent space, and the nearest neighbor classification may be used to rank the road sign templates. See also paragraph [0020, 0058].),
wherein the confidence level is calculated based on a distance between the obtained representative vector to other obtained representative vectors (Fig. 5, Paragraph [0055] – KIM discloses the system may integrate a machine learning model such as one-shot classifier to predict top-K road sign template candidates that are similar to a cropped image patch containing a road sign. The inputs for this model may be (1) a cropped image patch around the bounding box that the annotator draws on the real road image, and (2) sign templates filtered by the Road Sign Knowledge Graph. These two inputs above may be encoded into the latent space, and the nearest neighbor classification may be used to rank the road sign templates. See also paragraph [0020, 0058].).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date the claimed invention was made to combine the teachings of ZHAO in view of KIM of having a non-transitory computer readable medium for dynamic classification for autonomous driving, the non-transitory computer readable medium stores instructions for: producing, by a classification neural network associated with a driving of a vehicle, a catalog sample representative vector representing a catalog sample of a new class for autonomous driving being unidentified by the classification neural network at the time of the driving, generating, by the classification neural network, a new image-based representative vector representing an image of an entity of the new class; and classifying, by the classification neural network, the entity as being associated with the new class, with the teachings of KIM of having wherein the catalog sample is a diagram or graphical illustration of an item that differs from an actual image of the item; wherein the classification neural network is trained by a training process that comprises clustering image-based representative vectors of a given class based on distances between the image-based representative vectors of the given class and a catalog sample representative vector representing a catalog sample of the given class;
Wherein having ZHAO’s non-transitory computer readable medium for dynamic classification for autonomous driving wherein an obtained representative vector of an existing class is classified with a confidence level, wherein the confidence level is calculated based on a distance between the obtained representative vector to other obtained representative vectors.
The motivation behind the modification would have been to obtain an improved method of identifying new classes that enables efficient resource usage and reduces manual annotation efforts, since both ZHAO and KIM relate to image processing and classification using machine learning models, wherein ZHAO relates to dynamic class incremental learning without forgetting that may beneficially combine embedding representation learning plus index and search for incremental learning without forgetting on-device, as such, enabling efficient resource usage by reducing, and in some aspects, eliminating retraining from scratch at the arrival of new data, and KIM relates to machine learning and road sign recognition wherein utilizing a knowledge graph approach can reduce sign search space; the system can propose the correct single candidate by and large, reducing the human search effort fairly. Please see ZHAO (WO 2024197443 A1), Paragraph [0029, 0079], KIM (US 20220067405 A1), Paragraph [0020].
Regarding claim 16, ZHAO in view of KIM teach the non-transitory computer readable medium according to claim 11,
ZHAO further teaches wherein the training process comprises applying, for the given class, a loss process that is operative to reduce an angle between image-based representative vectors of the given class and a catalog sample representing vector of the given class (Fig. 5, Paragraph [0063] – ZHAO discloses input 502 may, for instance, include pairs of images or a batch of samples. The backbone 504 may process the input 502, extracting features of the input 502 to generate a representation of the input 502. Paragraph [0065] – ZHAO discloses the input 502 may be mined for pairs of samples (may also be referred to as "examples") and weighted for pair-based loss functions. The example architecture 500 may learn to minimize the distance (e.g., cosine distance [wherein cosine distance is a loss process that is operative to reduce an angle]) between similar samples and to maximize the distance between dissimilar samples.)
and increase the angle between the image-based representative vectors of the given class and other catalog sample representative vectors of other classes (Fig. 5, Paragraph [0063] – ZHAO discloses input 502 may, for instance, include pairs of images or a batch of samples. The backbone 504 may process the input 502, extracting features of the input 502 to generate a representation of the input 502. Paragraph [0065] – ZHAO discloses the input 502 may be mined for pairs of samples (may also be referred to as "examples") and weighted for pair-based loss functions. The example architecture 500 may learn to minimize the distance (e.g., cosine distance [wherein cosine distance is a loss process that is operative to increase the angle]) between similar samples and to maximize the distance between dissimilar samples.).
Regarding claim 17, ZHAO in view of KIM teach the non-transitory computer readable medium according to claim 11,
ZHAO fails to explicitly teach wherein the training process involves applying a triple loss process.
However, KIM explicitly teaches wherein the training process involves applying a triple loss process (Fig. 5, Paragraph [0058] – KIM discloses to improve the one-shot classifier, we introduce metric learning, such as triplet loss, during training to further separate different classes in the embedding space.).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date the claimed invention was made to combine the teachings of ZHAO in view of KIM of having a non-transitory computer readable medium for dynamic classification for autonomous driving, the non-transitory computer readable medium stores instructions for: producing, by a classification neural network associated with a driving of a vehicle, a catalog sample representative vector representing a catalog sample of a new class for autonomous driving being unidentified by the classification neural network at the time of the driving, wherein the classification neural network is trained by a training process that comprises clustering image-based representative vectors of a given class based on distances between the image-based representative vectors of the given class and a catalog sample representative vector representing a catalog sample of the given class; with the teachings of KIM of having wherein the training process involves applying a triple loss process.
Wherein having ZHAO’s non-transitory computer readable medium for dynamic classification for autonomous driving wherein the training process involves applying a triple loss process.
The motivation behind the modification would have been to obtain an improved method of identifying new classes that enables efficient resource usage and reduces manual annotation efforts, since both ZHAO and KIM relate to image processing and classification using machine learning models, wherein ZHAO relates to dynamic class incremental learning without forgetting that may beneficially combine embedding representation learning plus index and search for incremental learning without forgetting on-device, as such, enabling efficient resource usage by reducing, and in some aspects, eliminating retraining from scratch at the arrival of new data, and KIM relates to machine learning and road sign recognition wherein utilizing a knowledge graph approach can reduce sign search space; the system can propose the correct single candidate by and large, reducing the human search effort fairly. Please see ZHAO (WO 2024197443 A1), Paragraph [0029, 0079], KIM (US 20220067405 A1), Paragraph [0020].
Regarding claim 18, ZHAO in view of KIM teach the non-transitory computer readable medium according to claim 11,
ZHAO fails to explicitly teach wherein the producing comprises performing a one-shot learning process.
However, KIM explicitly teaches wherein the producing comprises performing a one-shot learning process (Fig. 5, Paragraph [0058] – KIM discloses next classifier 515 may be a one-shot classifier based on an encoder, such as a vibrational prototyping encoder, which learns a similarity metric to classify images at test time using a single prototype for each road sign within the candidate classes.).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date the claimed invention was made to combine the teachings of ZHAO in view of KIM of having a non-transitory computer readable medium for dynamic classification for autonomous driving, the non-transitory computer readable medium stores instructions for: producing, by a classification neural network associated with a driving of a vehicle, a catalog sample representative vector representing a catalog sample of a new class for autonomous driving being unidentified by the classification neural network at the time of the driving, with the teachings of KIM of having wherein the producing comprises performing a one-shot learning process.
Wherein having ZHAO’s non-transitory computer readable medium for dynamic classification for autonomous driving wherein the producing comprises performing a one-shot learning process.
The motivation behind the modification would have been to obtain an improved method of identifying new classes that enables efficient resource usage and reduces manual annotation efforts, since both ZHAO and KIM relate to image processing and classification using machine learning models, wherein ZHAO relates to dynamic class incremental learning without forgetting that may beneficially combine embedding representation learning plus index and search for incremental learning without forgetting on-device, as such, enabling efficient resource usage by reducing, and in some aspects, eliminating retraining from scratch at the arrival of new data, and KIM relates to machine learning and road sign recognition wherein utilizing a knowledge graph approach can reduce sign search space; the system can propose the correct single candidate by and large, reducing the human search effort fairly. Please see ZHAO (WO 2024197443 A1), Paragraph [0029, 0079], KIM (US 20220067405 A1), Paragraph [0020].
Regarding claim 19, ZHAO in view of KIM teach the non-transitory computer readable medium according to claim 11,
ZHAO fails to explicitly teach wherein the producing comprises performing a few-shot learning process.
However, KIM explicitly teaches wherein the producing comprises performing a few-shot learning process (Fig. 6C, Paragraph [0062] – KIM discloses The system may output road sign icons along with the image patch (cropped based on human annotator's input of a geometric shape) that are used for few shot learning classifier that produces top K candidates. The search space (# of candidates) may be reduced by the Knowledge Graph provides a better prediction in few shot learning classifier.).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date the claimed invention was made to combine the teachings of ZHAO in view of KIM of having a non-transitory computer readable medium for dynamic classification for autonomous driving, the non-transitory computer readable medium stores instructions for: producing, by a classification neural network associated with a driving of a vehicle, a catalog sample representative vector representing a catalog sample of a new class for autonomous driving being unidentified by the classification neural network at the time of the driving, with the teachings of KIM of having wherein the producing comprises performing a few-shot learning process.
Wherein having ZHAO’s non-transitory computer readable medium for dynamic classification for autonomous driving wherein the producing comprises performing a few-shot learning process.
The motivation behind the modification would have been to obtain an improved method of identifying new classes that enables efficient resource usage and reduces manual annotation efforts, since both ZHAO and KIM relate to image processing and classification using machine learning models, wherein ZHAO relates to dynamic class incremental learning without forgetting that may beneficially combine embedding representation learning plus index and search for incremental learning without forgetting on-device, as such, enabling efficient resource usage by reducing, and in some aspects, eliminating retraining from scratch at the arrival of new data, and KIM relates to machine learning and road sign recognition wherein utilizing a knowledge graph approach can reduce sign search space; the system can propose the correct single candidate by and large, reducing the human search effort fairly. Please see ZHAO (WO 2024197443 A1), Paragraph [0029, 0079], KIM (US 20220067405 A1), Paragraph [0020].
Regarding claim 20, ZHAO in view of KIM teach the non-transitory computer readable medium according to claim 11,
ZHAO fails to explicitly teach wherein the image of the entity of the new class is a cropped image of the entity of the new class.
However, KIM explicitly teaches wherein the image of the entity of the new class is a cropped image of the entity of the new class (Fig. 5, Paragraph [0055] – KIM discloses the system may integrate a machine learning model such as one-shot classifier to predict top-K road sign template candidates that are similar to a cropped image patch containing a road sign. The inputs for this model may be (1) a cropped image patch around the bounding box that the annotator draws on the real road image, and (2) sign templates filtered by the Road Sign Knowledge Graph. These two inputs above may be encoded into the latent space, and the nearest neighbor classification may be used to rank the road sign templates.).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date the claimed invention was made to combine the teachings of ZHAO in view of KIM of having a non-transitory computer readable medium for dynamic classification for autonomous driving, the non-transitory computer readable medium stores instructions for: producing, by a classification neural network associated with a driving of a vehicle, a catalog sample representative vector representing a catalog sample of a new class for autonomous driving being unidentified by the classification neural network at the time of the driving, with the teachings of KIM of having wherein the image of the entity of the new class is a cropped image of the entity of the new class.
Wherein having ZHAO’s non-transitory computer readable medium for dynamic classification for autonomous driving wherein the image of the entity of the new class is a cropped image of the entity of the new class.
The motivation behind the modification would have been to obtain an improved method of identifying new classes that enables efficient resource usage and reduces manual annotation efforts, since both ZHAO and KIM relate to image processing and classification using machine learning models, wherein ZHAO relates to dynamic class incremental learning without forgetting that may beneficially combine embedding representation learning plus index and search for incremental learning without forgetting on-device, as such, enabling efficient resource usage by reducing, and in some aspects, eliminating retraining from scratch at the arrival of new data, and KIM relates to machine learning and road sign recognition wherein utilizing a knowledge graph approach can reduce sign search space; the system can propose the correct single candidate by and large, reducing the human search effort fairly. Please see ZHAO (WO 2024197443 A1), Paragraph [0029, 0079], KIM (US 20220067405 A1), Paragraph [0020].
Claims 5 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over --ZHAO (WO 2024197443 A1), hereinafter referenced as ZHAO in view of KIM (US 20220067405 A1), hereinafter referenced as KIM in further view of HICKSON (US 20200027002 A1), hereinafter referenced as HICKSON.
Regarding claim 5, ZHAO in view of KIM teach the method according to claim 1,
Although ZHAO further teaches and increase the distance between the image-based representative vectors of the given class and other catalog sample representative vectors of other classes (Fig. 5, Paragraph [0063] – ZHAO discloses input 502 may, for instance, include pairs of images or a batch of samples. The backbone 504 may process the input 502, extracting features of the input 502 to generate a representation of the input 502. Paragraph [0065] – ZHAO discloses the input 502 may be mined for pairs of samples (may also be referred to as "examples") and weighted for pair-based loss functions. The example architecture 500 may learn to minimize the distance (e.g., cosine distance) between similar samples and to maximize the distance between dissimilar samples.).
ZHAO in view of KIM fail to explicitly teach wherein the training process comprises applying, for the given class, a loss process that is operative to reduce a distance between image-based representative vectors of the given class and a catalog sample representing vector of the given class.
However, HICKSON explicitly teaches wherein the training process (Fig. 1, Paragraph [0043]) comprises applying, for the given class, a loss process that is operative to reduce a distance between image-based representative vectors of the given class and a catalog sample representing vector of the given class (Fig. 1, Paragraph [0044] – HICKSON discloses the objective function includes a classification loss 132 and, when the selected image is an object image 106, a clustering loss 134. The clustering loss 134 encourages the embedding 128 to be similar (e.g., according to some numerical similarity measure) to at least one of the cluster centers 130.).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date the claimed invention was made to combine the teachings of ZHAO in view of KIM of having a method of dynamic classification for autonomous driving, the method comprises: producing, by a classification neural network associated with a driving of a vehicle, a catalog sample representative vector representing a catalog sample of a new class for autonomous driving being unidentified by the classification neural network at the time of the driving, generating, by the classification neural network, a new image-based representative vector representing an image of an entity of the new class; and classifying, by the classification neural network, the entity as being associated with the new class, with the teachings of KIM of having wherein the training process comprises applying, for the given class, a loss process that is operative to reduce a distance between image-based representative vectors of the given class and a catalog sample representing vector of the given class.
Wherein having ZHAO’s method of dynamic classification for autonomous driving wherein the training process comprises applying, for the given class, a loss process that is operative to reduce a distance between image-based representative vectors of the given class and a catalog sample representing vector of the given class.
The motivation behind the modification would have been to obtain an improved method of identifying new classes that enables efficient resource usage, since both ZHAO and HICKSON relate to image processing using machine learning models, wherein ZHAO relates to dynamic class incremental learning without forgetting that may beneficially combine embedding representation learning plus index and search for incremental learning without forgetting on-device, as such, enabling efficient resource usage by reducing, and in some aspects, eliminating retraining from scratch at the arrival of new data, and HICKSON relates to methods, systems, and apparatus for determining a clustering of images into a plurality of semantic categories; moreover, the system may consume fewer computational resources than some conventional systems (e.g., memory and computing power) than some conventional systems since jointly performing image embedding and image clustering enables these tasks to be performed more efficiently (e.g., over fewer iterations) than if they were performed separately. Please see ZHAO (WO 2024197443 A1), Paragraph [0029, 0079], HICKSON (US 20200027002 A1), Paragraph [0022].
Regarding claim 15, ZHAO in view of HICKSON teach the non-transitory computer readable medium according to claim 14,
Although ZHAO further teaches and increase the distance between the image-based representative vectors of the given class and other catalog sample representative vectors of other classes (Fig. 5, Paragraph [0063] – ZHAO discloses input 502 may, for instance, include pairs of images or a batch of samples. The backbone 504 may process the input 502, extracting features of the input 502 to generate a representation of the input 502. Paragraph [0065] – ZHAO discloses the input 502 may be mined for pairs of samples (may also be referred to as "examples") and weighted for pair-based loss functions. The example architecture 500 may learn to minimize the distance (e.g., cosine distance) between similar samples and to maximize the distance between dissimilar samples.),
ZHAO in view of KIM fail to explicitly teach wherein the training process comprises applying, for the given class, a loss process that is operative to reduce a distance between image-based representative vectors of the given class and a catalog sample representing vector of the given class.
However, HICKSON explicitly teaches wherein the training process (Fig. 1, Paragraph [0043]) comprises applying, for the given class, a loss process that is operative to reduce a distance between image-based representative vectors of the given class and a catalog sample representing vector of the given class (Fig. 1, Paragraph [0044] – HICKSON discloses the objective function includes a classification loss 132 and, when the selected image is an object image 106, a clustering loss 134. The clustering loss 134 encourages the embedding 128 to be similar (e.g., according to some numerical similarity measure) to at least one of the cluster centers 130.).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date the claimed invention was made to combine the teachings of ZHAO in view of KIM of having a non-transitory computer readable medium for dynamic classification for autonomous driving, the non-transitory computer readable medium stores instructions for: producing, by a classification neural network associated with a driving of a vehicle, a catalog sample representative vector representing a catalog sample of a new class for autonomous driving being unidentified by the classification neural network at the time of the driving, generating, by the classification neural network, a new image-based representative vector representing an image of an entity of the new class; and classifying, by the classification neural network, the entity as being associated with the new class, with the teachings of KIM of having wherein the training process comprises applying, for the given class, a loss process that is operative to reduce a distance between image-based representative vectors of the given class and a catalog sample representing vector of the given class.
Wherein having ZHAO’s non-transitory computer readable medium for dynamic classification for autonomous driving wherein the training process comprises applying, for the given class, a loss process that is operative to reduce a distance between image-based representative vectors of the given class and a catalog sample representing vector of the given class.
The motivation behind the modification would have been to obtain an improved method of identifying new classes that enables efficient resource usage, since both ZHAO and HICKSON relate to image processing using machine learning models, wherein ZHAO relates to dynamic class incremental learning without forgetting that may beneficially combine embedding representation learning plus index and search for incremental learning without forgetting on-device, as such, enabling efficient resource usage by reducing, and in some aspects, eliminating retraining from scratch at the arrival of new data, and HICKSON relates to methods, systems, and apparatus for determining a clustering of images into a plurality of semantic categories; moreover, the system may consume fewer computational resources than some conventional systems (e.g., memory and computing power) than some conventional systems since jointly performing image embedding and image clustering enables these tasks to be performed more efficiently (e.g., over fewer iterations) than if they were performed separately. Please see ZHAO (WO 2024197443 A1), Paragraph [0029, 0079], HICKSON (US 20200027002 A1), Paragraph [0022].
Conclusion
Listed below are the prior arts made of record and not relied upon but are considered pertinent to applicant’s disclosure.
Fichman et al. (US 20240312046 A1) - System and methods for visually tracking movement associated with a mobile device in order to determine optimal locations for handing over the device between different base stations arranged in a sequence. Initially, a current location of the mobile device is determined via a radio access network associated with the device. Then, a moving object is visually detected in an immediate proximity to the location determined, using a visual sensor array associated with the radio access network. The moving object detected is assumed to be associated with the mobile device due to said proximity, and is hence visually tracked along its path of progression, until arriving at a location that was previously determined to be optimal for performing a handover between two adjacent base stations. A handover is then executed, while continuing with said visual tracking until another handover is required.… Fig. 1A-B, Abstract.
Utasi et al. (US 20230298181 A1) - The invention is a method for object segmentation in an image, comprising the steps of inputting the image to a trained machine learning system, and reconstructing the segmentation contour of the object. The method is characterized by comprising the steps of estimating, by the trained machine learning system, a representation of a segmentation contour of an object in the image, wherein the segmentation contour is a closed two-dimensional parametric curve, each point of which is defined by two coordinate components, wherein both coordinate components are parametrized.… Fig. 1, Abstract.
Bisain et al. (US 20220198677 A1) - Examples are described for processing images to mask dynamic objects out of images to improve feature tracking between images. A device receives an image of an environment captured by an image sensor. The image depicts at least a static portion of the environment and a dynamic object in the environment. The device identifies a portion of the image that includes a depiction of the dynamic object. For example, the device can detect a bounding box around the dynamic object, or can detect which pixels in the image correspond to the dynamic object. The device generates a masked image at least by masking the portion of the image. The device identifies features in the masked image, and uses the features from the masked image for feature tracking from other images of the environment.… Fig. 1, Abstract.
Burlina et al. (US 20250076486 A1) - Vehicle computing systems may receive and fuse long-wave infrared data with radar data to detect and track objects in low-visibility driving environments. In some examples, radar data including position and velocity data may be projected over infrared image data to detect, classify, and track infrared-emitting objects. Machine-learned transformer models with attention also may be trained to output object detections based on combined infrared and radar data. The fusion of infrared and radar data may be used individual and/or may be synchronized with other sensor modalities. In some examples, the fusion and analysis of infrared and radar data may be used in specific low-visibility driving environments… Fig. 1, Abstract.
Aljundi et al. (EP 4432163 A1) - A computer-implemented method of generating a machine learning model for classification using a pre-trained backbone model, comprising:- for a first training session, training (E30) the pre-trained backbone model using a first dataset,- connecting (E50) a classification head model on top of the trained backbone model, so as to obtain a main model,- for at least one second training session, training (E70) the main model using a second dataset while keeping the parameters of the trained backbone model fixed, the second dataset comprising at least one class distinct from the classes of the first dataset.… Fig. 1, Abstract.
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any extension fee pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to BEZAWIT N SHIMELES whose telephone number is (571)272-7663. The examiner can normally be reached M-F 7:30am-5pm.
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, Chineyere Wills-Burns can be reached at (571) 272-9752. 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.
/BEZAWIT NOLAWI SHIMELES/Examiner, Art Unit 2673
/CHINEYERE WILLS-BURNS/Supervisory Patent Examiner, Art Unit 2673
Prosecution Insights