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 .
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submissions, filed on May 26th, 2026, have been entered.
Status of Claims
Claims 1-20 are pending, claims 1, 9 and 17 have been amended. Claims 1-20 remains rejected.
Response to Argument(s)
101 rejection:
In view of the Amendments to independent claims 1, 9 and 17 the previously applied 101 rejections are withdrawn. Applicants’ arguments found to be persuasive, the 101 rejection has been withdrawn.
rejections:
In view of the Amendments to independent claims 1, 9 and 17 the previously applied prior art rejections are withdrawn. Applicants’ arguments are rendered moot in view of the new grounds of rejection set forth below.
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.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1-20 are rejected under 35 U.S.C. 103 as being unpatentable over Faisal Mehmood et. al. (“Object Detection Mechanism Based on Deep Learning Algorithm using Embedded IoT Devices for Smart Home Appliances Control in CoT, March 2019, Journal of Ambient Intelligence and Humanized Computing” hereinafter as “Mehmood”) in view of Ting-I Hsieh et. al. (“One-Shot Object Detection with Co-Attention and Co-Excitation, Sept. 2019, NeurIPS 2019, CMT Num 1562” hereinafter as ”Hsieh”).
Regarding claim 1, Mehmood discloses a computer-implemented method for loading a first machine learning model into a camera system to detect a first feature (the system such as shown in FIG. 1 of the camera system using smart home management service and smart control service), comprising: receiving, by at least one computer processor on the camera system, a command to download the first machine learning model to the camera system, wherein the first machine learning model is configured to detect a first feature in a video stream (section 3, 1st par., discloses the service can be used on a phone through a phone app which includes wherein live streaming video of the environment for control, the user can be understood to download the app on the phone [a command to download] wherein the app includes a machine learning model such as shown in FIG. 4 to detect a feature in a video stream, by BRI, covers the scope of the claim); downloading the first machine learning model to the camera system (downloading the app indicates downloading the machine learning model to the camera system including the phone app.); installing the first machine learning model on the camera system (the downloading would install the machine learning model onto the camera system, by BRI, covers the scope of the claim); capturing a video stream (the live streaming of the environment, as discussed previously, indicates capturing a video stream); detecting, using the first machine learning model, an unknown feature in the video stream (the machine learning model of Fig. 4 is to detect the feature in the video stream; the term “unknown features” can be interpreted to be any features that are being processed or detected from the image since, they are to be determined/to be known of more processed information/data, such as shown in figure 4 wherein the features maps [first features] and the default bounding box to be analogous to the unknown feature since they are to be determined to be known and processed to be specific rather than default); in response to the detecting, providing, via a network, one or more frames of the video stream including the unknown feature to a server (figure 4 discloses a Single Shot Detector, which is analogous to second machine learning as claimed, since the SSD is a deep neural network according to section 2, last par., moreover, the input to the SSD includes the information previously proceed including the already mapped “unknown features,” “one or more frames” of the preprocessing step, since the Single Shot Detector processing is sequential to the previous processing therefore, it’s analogous to “in response to the detecting,…”) comprising a second machine learning model configured to detect second features in addition to the first feature (the Single Shot Detector, as discussed previously and shown in figure 4, the Single Shot Detector [analogous to the 2nd machine learning] to process of multibox being bounding box regression [analogous to the recited second features] which is in addition to the feature maps and the default bounding box, since these information/data are being fed into the Single Shot Detector, therefore, all of these information/data are used together for the object localization result [analogous to the recited “second features in addition to the first feature”]); receiving, via the network, a classification label corresponding to the unknown feature from the server (as shown in figure 4, the output to the processing is object classification which is analogous to the classification label as claimed, of the process data/information, as previously mapped to be the unknown feature), and in response to receiving the classification label, transmitting a camera detection notification indicating detection of the feature to a user device corresponding to a user account linked to the camera system (section 3, 1st par., discloses the user can communicate with the system through phone hence, it indicates that, the user can observe the live stream with the detected objects in the stream to send control command through the user account linked to the camera system such as shown in FIG. 5; which is in response to the step and the output of the figure 4 hence , is analogous to “in response to receiving, the classification label” as claimed).
However, Mehmood does not explicitly teach wherein the unknown feature is not classifiable by the first machine learning model; and classify the unknown feature, wherein the second machine learning model is separate from the first machine learning model.
In the same field of One-Shot Object Detection (Title, Hsieh), Hsieh explicitly teaches wherein the unknown feature is not classifiable by the first machine learning model (section 4, 1st paragraph, discloses “we train and evaluate our model on VOC and COCO benchmark datasets….splits of seen and unseen VOC classes…alternately taking three splits as seen classes and one split as unseen classes” therefore, the model trained using the COCO dataset is analogous to Mehmood’s pretrained model on COCO dataset, moreover, Hsieh teaches that the features being processed include seen and unseen classes being split, therefore, for the seen classes being processed by the model, the unseen features are not being processed by the same model, therefore, it can be understood that the unseen features classes here pertain the unknown features as claimed, as for the processing of the model on the seen classes, these unseen classes are then, at the same instance, would not be processed or classified, in other words they are unclassifiable by the same model, hence the scope of the claim falls under the same scope of Hsieh’s teaching); and classify the unknown feature, wherein the second machine learning model is separate from the first machine learning model (Page 2, “Few-Shot Object Detection” section, discloses the few-shot classification, which is analogous to Mehmood’s single-shot detector, being mapped to the second machine learning model, moreover, the section discloses “training on a handful of labeled images of unseen classes” which uses such model to classify images with unseen classes of the unknown features, therefore, the 2nd machine learning here is able to classify the images with the unseen classes or the unknown features, which is also integrated into the current invention’s model, in Page 5, “ImageNet pre-training” section, discloses “to ensure that our model does not foresee the unseen classes” therefore, the model that performs on the unseen class split here is able to classify the unknown features as claimed).
Thus, it would have been obvious for a person of ordinary skill in the art before the effective filing date to modify Mehmood’s method of obtaining dataset to be processed by first machine learning model with unknown features extracted from these dataset, to be forward to a second machine learning model for further processing;
Wherein Mehmood’s method can be modified to be have the unknown feature is not classifiable by the first machine learning model; and classify the unknown feature, wherein the second machine learning model is separate from the first machine learning model as taught by Hsieh as discussed in the mapping above.
Such a modification is the result of combing prior art elements according to known methods to yield predictable results. The motivation for the proposed modification would have been to tackle problem in machine classification, to label seen and unseen classes in training more accurately using such method involving one-shot object detection, so that the object detection can be performed more accurately with accurate classes (Abstract, Hsieh).
Regarding claim 2, Mehmood in view of Hsieh discloses the computer-implemented method of claim 1, Hsieh further explicitly teaches comprising: setting a response command to perform a home automation action in response to detection of the unknown feature in the video stream (through the detection of Fig. 5, the user can control the home appliances such as shown in FIG. 6 therefore, is analogous to setting a response command to perform a home automation action [controlling the appliances] in response to the result of the detection of FIG. 5; therefore, by BRI, covers the scope of the claim); and transmitting, to a home automation system, the response command to perform the home automation action (then send to home automation system to perform the home automation action to control the appliance such as shown in FIG 6 and FIG. 8).
Regarding claim 3, Mehmood in view of Hsieh discloses the computer-implemented method of claim 1, wherein Mehmood teaches he installing further comprises: retraining the machine learning model using one or more images captured by the camera system, thereby modifying one or more parameters used by the first machine learning model to detect the feature (page 7, 1st par., discloses a pre-trained model is used for the classifying objects and then perform the training of the pre-trained model is further discloses by modifying the model to teach the network to detect objects of various sizes based on IoU ratios [modifying the parameters as claimed, by BRI], which indicates a retraining, by BRI, covers the scope of the claim).
Regarding claim 4, Mehmood in view of Hsieh discloses the computer-implemented method of claim 1, wherein Mehmood teaches further comprising: retraining the first machine learning model using the classification label to classify the unknown feature (page 7, 1st par., discloses a pre-trained model is used for the classifying objects and then perform the training of the pre-trained model is further discloses by modifying the model to teach the network to detect objects of various sizes based on IoU ratios [modifying the parameters as claimed, by BRI], which indicates a retraining, by BRI, covers the scope of the claim).
Regarding claim 5, Mehmood in view of Hsieh discloses the computer-implemented method of claim 1, wherein Mehmood teaches further comprising: receiving data from a second camera system indicating detection of a third feature by a third machine learning model installed on the second camera system (FIG. 6 shows that the home environment can include several areas such as bedroom or other room to be entered the room name to access, therefore, it indicates the use of more than one camera for the detection of the corresponding object [third feature as claimed, by BRI], the machine learning model used for detecting of the corresponding room can be understood to be the third machine learning model and the corresponding cameras to be the second camera system, by BRI, covers the scope of the claim); and transmitting a second response command to perform a second home automation action corresponding to detection of the first feature in the video stream and the third feature (the processing of the second camera is the same which includes transmitting a second response to command a perform of a second home automation action corresponding to the detection result, such as for the analogous limitation in claim 1 above).
Regarding claim 6, Mehmood in view of Hsieh discloses the computer-implemented method of claim 1, wherein Mehmood teaches the first feature is an appearance of a predefined object in the video stream (as discussed above in claim 1, the feature is an appearance of a predefined object in the room to be live streamed such as a light in a bedroom as shown in FIG. 6).
Regarding claim 7, Mehmood in view of Hsieh discloses the computer-implemented method of claim 1, wherein Mehmood teaches the first feature is an absence of a previously detected object in the video stream (such as shown in FIG. 4 and disclosed in page 7, the system can detect new objects there were not previously classified, in this instance, the feature would be new object which is not the same objects being previously identified, in other words, an absence of the previously detected object in the video stream being the feature as discussed, by BRI, covers the scope of the claim).
Regarding claim 8, Mehmood in view of Hsieh discloses the computer-implemented method of claim 1, wherein Mehmood teaches the first feature is recognition of a semantic meaning corresponding to textual characters identified via natural language processing (section 4., 1st par., and FIG. 6 shows that the processing can be the processing of SMS message to control the device which includes textual characters to be processed into command, the machine learning to process so can be understood to be the natural language processing, by BRI, covers the scope of the claim).
Regarding claim 9, Mehmood discloses a camera system, comprising: one or more cameras; one or more memories; and at least one processor each coupled to the one or more cameras and at least one of the memories and configured to perform operations (the system such as shown in FIG. 1 of the camera system using smart home management service and smart control service, using of a computer with computer components), comprising: receiving, by at least one computer processor on the camera system, a command to download the first machine learning model to the camera system, wherein the first machine learning model is configured to detect a first feature in a video stream (section 3, 1st par., discloses the service can be used on a phone through a phone app which includes wherein live streaming video of the environment for control, the user can be understood to download the app on the phone [a command to download] wherein the app includes a machine learning model such as shown in FIG. 4 to detect a feature in a video stream, by BRI, covers the scope of the claim); downloading the first machine learning model to the camera system (downloading the app indicates downloading the machine learning model to the camera system including the phone app.); installing the first machine learning model on the camera system (the downloading would install the machine learning model onto the camera system, by BRI, covers the scope of the claim); capturing a video stream (the live streaming of the environment, as discussed previously, indicates capturing a video stream); detecting, using the first machine learning model, an unknown feature in the video stream (the machine learning model of Fig. 4 is to detect the feature in the video stream; the term “unknown features” can be interpreted to be any features that are being processed or detected from the image since, they are to be determined/to be known of more processed information/data, such as shown in figure 4 wherein the features maps [first features] and the default bounding box to be analogous to the unknown feature since they are to be determined to be known and processed to be specific rather than default); in response to the detecting, providing, via a network, one or more frames of the video stream including the unknown feature to a server (figure 4 discloses a Single Shot Detector, which is analogous to second machine learning as claimed, since the SSD is a deep neural network according to section 2, last par., moreover, the input to the SSD includes the information previously proceed including the already mapped “unknown features,” “one or more frames” of the preprocessing step, since the Single Shot Detector processing is sequential to the previous processing therefore, it’s analogous to “in response to the detecting,…”) comprising a second machine learning model configured to detect second features in addition to the first feature (the Single Shot Detector, as discussed previously and shown in figure 4, the Single Shot Detector [analogous to the 2nd machine learning] to process of multibox being bounding box regression [analogous to the recited second features] which is in addition to the feature maps and the default bounding box, since these information/data are being fed into the Single Shot Detector, therefore, all of these information/data are used together for the object localization result [analogous to the recited “second features in addition to the first feature”]); receiving, via the network, a classification label corresponding to the unknown feature from the server (as shown in figure 4, the output to the processing is object classification which is analogous to the classification label as claimed, of the process data/information, as previously mapped to be the unknown feature), and in response to receiving the classification label, transmitting a camera detection notification indicating detection of the feature to a user device corresponding to a user account linked to the camera system (section 3, 1st par., discloses the user can communicate with the system through phone hence, it indicates that, the user can observe the live stream with the detected objects in the stream to send control command through the user account linked to the camera system such as shown in FIG. 5; which is in response to the step and the output of the figure 4 hence , is analogous to “in response to receiving, the classification label” as claimed).
However, Mehmood does not explicitly teach wherein the unknown feature is not classifiable by the first machine learning model; and classify the unknown feature, wherein the second machine learning model is separate from the first machine learning model.
In the same field of One-Shot Object Detection (Title, Hsieh), Hsieh discloses wherein the unknown feature is not classifiable by the first machine learning model (section 4, 1st paragraph, discloses “we train and evaluate our model on VOC and COCO benchmark datasets….splits of seen and unseen VOC classes…alternately taking three splits as seen classes and one split as unseen classes” therefore, the model trained using the COCO dataset is analogous to Mehmood’s pretrained model on COCO dataset, moreover, Hsieh teaches that the features being processed include seen and unseen classes being split, therefore, for the seen classes being processed by the model, the unseen features are not being processed by the same model, therefore, it can be understood that the unseen features classes here pertain the unknown features as claimed, as for the processing of the model on the seen classes, these unseen classes are then, at the same instance, would not be processed or classified, in other words they are unclassifiable by the same model, hence the scope of the claim falls under the same scope of Hsieh’s teaching); and classify the unknown feature, wherein the second machine learning model is separate from the first machine learning model (Page 2, “Few-Shot Object Detection” section, discloses the few-shot classification, which is analogous to Mehmood’s single-shot detector, being mapped to the second machine learning model, moreover, the section discloses “training on a handful of labeled images of unseen classes” which uses such model to classify images with unseen classes of the unknown features, therefore, the 2nd machine learning here is able to classify the images with the unseen classes or the unknown features, which is also integrated into the current invention’s model, in Page 5, “ImageNet pre-training” section, discloses “to ensure that our model does not foresee the unseen classes” therefore, the model that performs on the unseen class split here is able to classify the unknown features as claimed).
Thus, it would have been obvious for a person of ordinary skill in the art before the effective filing date to modify Mehmood’s method of obtaining dataset to be processed by first machine learning model with unknown features extracted from these dataset, to be forward to a second machine learning model for further processing;
Wherein Mehmood’s method can be modified to be have the unknown feature is not classifiable by the first machine learning model; and classify the unknown feature, wherein the second machine learning model is separate from the first machine learning model as taught by Hsieh as discussed in the mapping above.
Such a modification is the result of combing prior art elements according to known methods to yield predictable results. The motivation for the proposed modification would have been to tackle problem in machine classification, to label seen and unseen classes in training more accurately using such method involving one-shot object detection, so that the object detection can be performed more accurately with accurate classes (Abstract, Hsieh).
Regarding claim 10, Mehmood in view of Hsieh discloses the camera system of claim 9, wherein Mehmood teaches the operations further comprising: setting a response command to perform a home automation action in response to detection of the feature in the video stream (through the detection of Fig. 5, the user can control the home appliances such as shown in FIG. 6 therefore, is analogous to setting a response command to perform a home automation action [controlling the appliances] in response to the result of the detection of FIG. 5; therefore, by BRI, covers the scope of the claim); and transmitting, to a home automation system, the response command to perform the home automation action (then send to home automation system to perform the home automation action to control the appliance such as shown in FIG 6 and FIG. 8).
Regarding claim 11, Mehmood in view of Hsieh discloses the camera system of claim 9, wherein Mehmood teaches the installing further comprises: retraining the machine learning model using one or more images captured by the camera system, thereby modifying one or more parameters used by the machine learning model to detect the feature (page 7, 1st par., discloses a pre-trained model is used for the classifying objects and then perform the training of the pre-trained model is further discloses by modifying the model to teach the network to detect objects of various sizes based on IoU ratios [modifying the parameters as claimed, by BRI], which indicates a retraining, by BRI, covers the scope of the claim).
Regarding claim 12, Mehmood in view of Hsieh discloses the camera system of claim 9, wherein Mehmood teaches the operations further comprising: presenting one or more frames of the video stream including the unknown feature to a system external to the camera system for classification of the unknown feature (page 7, 1st par., discloses the model can classify objects of unknown objects [unknown feature as claimed, by BRI] of the live streaming video of FIG. 4 and Fig. 5, to aa single shot detector [external system to the camera system for classification of the unknown feature as claimed, by BRI]); receiving a classification label corresponding to the unknown feature from the system external to the camera system (the system, as discussed previously, would output classification result of the object to the camera system as discussed); transmitting, to the user device, a second camera detection notification including the classification label (the camera system, as discussed previously, includes the user phone system to receive the notification including the class for the object classified such as shown in FIG. 4 and FIG. 5); and retraining the machine learning model using the classification label to classify the unknown feature (page 7, 1st par., discloses a pre-trained model is used for the classifying objects and then perform the training of the pre-trained model is further discloses by modifying the model to teach the network to detect objects of various sizes based on IoU ratios [modifying the parameters as claimed, by BRI], which indicates a retraining, by BRI, covers the scope of the claim).
Regarding claim 13, Mehmood in view of Hsieh discloses The camera system of claim 9, wherein Mehmood teaches the operations further comprising: receiving data from a second camera system indicating detection of a third feature by a third machine learning model installed on the second camera system (FIG. 6 shows that the home environment can include several areas such as bedroom or other room to be entered the room name to access, therefore, it indicates the use of more than one camera for the detection of the corresponding object [third feature as claimed, by BRI], the machine learning model used for detecting of the corresponding room can be understood to be the third machine learning model and the corresponding cameras to be the second camera system, by BRI, covers the scope of the claim); and transmitting a second response command to perform a second home automation action corresponding to detection of the first feature in the video stream and the third feature (the processing of the second camera is the same which includes transmitting a second response to command a perform of a second home automation action corresponding to the detection result, such as for the analogous limitation in claim 9 above).
Regarding claim 14, Mehmood in view of Hsieh discloses The camera system of claim 9, wherein Mehmood teaches wherein the first feature is an appearance of a predefined object in the video stream (as discussed above in claim 9, the feature is an appearance of a predefined object in the room to be live streamed such as a light in a bedroom as shown in FIG. 6).
Regarding claim 15, Mehmood in view of Hsieh discloses The camera system of claim 9, wherein Mehmood teaches the first feature is an absence of a previously detected object in the video stream (such as shown in FIG. 4 and disclosed in page 7, the system can detect new objects there were not previously classified, in this instance, the feature would be new object which is not the same objects being previously identified, in other words, an absence of the previously detected object in the video stream being the feature as discussed, by BRI, covers the scope of the claim).
Regarding claim 16, Mehmood in view of Hsieh discloses The camera system of claim 9, wherein Mehmood teaches the first feature is recognition of a semantic meaning corresponding to textual characters identified via natural language processing (section 4., 1st par., and FIG. 6 shows that the processing can be the processing of SMS message to control the device which includes textual characters to be processed into command, the machine learning to process so can be understood to be the natural language processing, by BRI, covers the scope of the claim).
Regarding claim 17, Mehmood discloses A non-transitory computer-readable medium having instructions stored thereon that, when executed by at least one computing device, cause the at least one computing device to perform operations (the system such as shown in FIG. 1 of the camera system using smart home management service and smart control service, using of a computer with computer components), comprising: receiving, by at least one computer processor on the camera system, a command to download the first machine learning model to the camera system, wherein the first machine learning model is configured to detect a first feature in a video stream (section 3, 1st par., discloses the service can be used on a phone through a phone app which includes wherein live streaming video of the environment for control, the user can be understood to download the app on the phone [a command to download] wherein the app includes a machine learning model such as shown in FIG. 4 to detect a feature in a video stream, by BRI, covers the scope of the claim); downloading the first machine learning model to the camera system (downloading the app indicates downloading the machine learning model to the camera system including the phone app.); installing the first machine learning model on the camera system (the downloading would install the machine learning model onto the camera system, by BRI, covers the scope of the claim); capturing a video stream (the live streaming of the environment, as discussed previously, indicates capturing a video stream); detecting, using the first machine learning model, an unknown feature in the video stream (the machine learning model of Fig. 4 is to detect the feature in the video stream; the term “unknown features” can be interpreted to be any features that are being processed or detected from the image since, they are to be determined/to be known of more processed information/data, such as shown in figure 4 wherein the features maps [first features] and the default bounding box to be analogous to the unknown feature since they are to be determined to be known and processed to be specific rather than default); in response to the detecting, providing, via a network, one or more frames of the video stream including the unknown feature to a server (figure 4 discloses a Single Shot Detector, which is analogous to second machine learning as claimed, since the SSD is a deep neural network according to section 2, last par., moreover, the input to the SSD includes the information previously proceed including the already mapped “unknown features,” “one or more frames” of the preprocessing step, since the Single Shot Detector processing is sequential to the previous processing therefore, it’s analogous to “in response to the detecting,…”) comprising a second machine learning model configured to detect second features in addition to the first feature (the Single Shot Detector, as discussed previously and shown in figure 4, the Single Shot Detector [analogous to the 2nd machine learning] to process of multibox being bounding box regression [analogous to the recited second features] which is in addition to the feature maps and the default bounding box, since these information/data are being fed into the Single Shot Detector, therefore, all of these information/data are used together for the object localization result [analogous to the recited “second features in addition to the first feature”]); receiving, via the network, a classification label corresponding to the unknown feature from the server (as shown in figure 4, the output to the processing is object classification which is analogous to the classification label as claimed, of the process data/information, as previously mapped to be the unknown feature), and in response to receiving the classification label, transmitting a camera detection notification indicating detection of the feature to a user device corresponding to a user account linked to the camera system (section 3, 1st par., discloses the user can communicate with the system through phone hence, it indicates that, the user can observe the live stream with the detected objects in the stream to send control command through the user account linked to the camera system such as shown in FIG. 5; which is in response to the step and the output of the figure 4 hence , is analogous to “in response to receiving, the classification label” as claimed).
However, Mehmood does not explicitly teach wherein the unknown feature is not classifiable by the first machine learning model; and classify the unknown feature, wherein the second machine learning model is separate from the first machine learning model.
In the same field of One-Shot Object Detection (Title, Hsieh), Hsieh discloses wherein the unknown feature is not classifiable by the first machine learning model (section 4, 1st paragraph, discloses “we train and evaluate our model on VOC and COCO benchmark datasets….splits of seen and unseen VOC classes…alternately taking three splits as seen classes and one split as unseen classes” therefore, the model trained using the COCO dataset is analogous to Mehmood’s pretrained model on COCO dataset, moreover, Hsieh teaches that the features being processed include seen and unseen classes being split, therefore, for the seen classes being processed by the model, the unseen features are not being processed by the same model, therefore, it can be understood that the unseen features classes here pertain the unknown features as claimed, as for the processing of the model on the seen classes, these unseen classes are then, at the same instance, would not be processed or classified, in other words they are unclassifiable by the same model, hence the scope of the claim falls under the same scope of Hsieh’s teaching); and classify the unknown feature, wherein the second machine learning model is separate from the first machine learning model (Page 2, “Few-Shot Object Detection” section, discloses the few-shot classification, which is analogous to Mehmood’s single-shot detector, being mapped to the second machine learning model, moreover, the section discloses “training on a handful of labeled images of unseen classes” which uses such model to classify images with unseen classes of the unknown features, therefore, the 2nd machine learning here is able to classify the images with the unseen classes or the unknown features, which is also integrated into the current invention’s model, in Page 5, “ImageNet pre-training” section, discloses “to ensure that our model does not foresee the unseen classes” therefore, the model that performs on the unseen class split here is able to classify the unknown features as claimed).
Thus, it would have been obvious for a person of ordinary skill in the art before the effective filing date to modify Mehmood’s method of obtaining dataset to be processed by first machine learning model with unknown features extracted from these dataset, to be forward to a second machine learning model for further processing;
Wherein Mehmood’s method can be modified to be have the unknown feature is not classifiable by the first machine learning model; and classify the unknown feature, wherein the second machine learning model is separate from the first machine learning model as taught by Hsieh as discussed in the mapping above.
Such a modification is the result of combing prior art elements according to known methods to yield predictable results. The motivation for the proposed modification would have been to tackle problem in machine classification, to label seen and unseen classes in training more accurately using such method involving one-shot object detection, so that the object detection can be performed more accurately with accurate classes (Abstract, Hsieh).
Regarding claim 18, Mehmood in view of Hsieh discloses t The non-transitory computer-readable medium of claim 17, wherein Mehmood teaches the operations further comprising: setting a response command to perform a home automation action in response to detection of the unknown feature in the video stream (through the detection of Fig. 5, the user can control the home appliances such as shown in FIG. 6 therefore, is analogous to setting a response command to perform a home automation action [controlling the appliances] in response to the result of the detection of FIG. 5; therefore, by BRI, covers the scope of the claim); and transmitting, to a home automation system, the response command to perform the home automation action (then send to home automation system to perform the home automation action to control the appliance such as shown in FIG 6 and FIG. 8).
Regarding claim 19, Mehmood in view of Hsieh discloses the camera system of claim 9, wherein Mehmood teaches the installing further comprises: retraining the first machine learning model using one or more images captured by the camera system, thereby modifying one or more parameters used by the machine learning model to detect the feature (page 7, 1st par., discloses a pre-trained model is used for the classifying objects and then perform the training of the pre-trained model is further discloses by modifying the model to teach the network to detect objects of various sizes based on IoU ratios [modifying the parameters as claimed, by BRI], which indicates a retraining, by BRI, covers the scope of the claim).
Regarding claim 20, Mehmood in view of Hsieh discloses the non-transitory computer-readable medium of claim 17, wherein Mehmood teaches the operations further comprising: retraining the first machine learning model using the classification label to classify the unknown feature (page 7, 1st par., discloses a pre-trained model is used for the classifying objects and then perform the training of the pre-trained model is further discloses by modifying the model to teach the network to detect objects of various sizes based on IoU ratios [modifying the parameters as claimed, by BRI], which indicates a retraining, by BRI, covers the scope of the claim).
Pertinent Prior Art(s)
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure:
Greene; Gregory et. al. , US 20180061158 A1, discloses determining and setting an authorization level for an unknown person are presented. Facial recognition may be performed on a received video stream. An authorized user may be recognized in the video feed. Additionally, an unknown person may be identified in the video feed. A provisional authorization level may be granted for the unknown person based on proximity between the unknown person and the authorized user in the received video stream.
JEONG; Gyu Taek et. al., US 20190028290 A1, discloses a face recognition device configured to compare a photographed authentication image of a user's face with a plurality of previously stored face images to perform authentication for user entering a house, a face recognition server configured to receive authenticated user information from the face recognition device and transmit access information for informing whether or not the user enters the house to a user terminal corresponding to the user information, and a home electronic device control unit configured to control the home electronic devices according to preset reservation information from a user when activated by the home automation activation unit, thereby operating the home electronic devices to correspond to the reservation information set by the user according to the user's access.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to PHUONG HAU CAI whose telephone number is (571)272-9424. The examiner can normally be reached M-F 8:30 am - 5:00pm.
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.
/PHUONG HAU CAI/Examiner, Art Unit 2673
/CHINEYERE WILLS-BURNS/Supervisory Patent Examiner, Art Unit 2673
Prosecution Insights