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 Arguments
Applicant’s arguments filed 6/9/26 with respect to the rejection of independent claims 1, 11, and 21 over Shmelkov in view of Li have been considered but are moot because the new ground of rejection. Please see the rejection below including Perez-Rua for the amended limitations directed to a plurality of feature extraction layers that construct feature maps provided to both the pretrained detection branch and the trainable detection branch operating in parallel (section 3.1-3.3).
Regarding claim 10, Zang is relied upon only for teaching an unmanned vehicle that tracks an object in accordance with user instructions. The communication of object detection information to a control system for training and the receipt of updated learned parameters are taught by Li and Shmelkov as set forth in the rejection of claim 1.
Please see the final rejection 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.
Claim(s) 1-8, 11-21 are rejected under 35 U.S.C. 103 as being unpatentable over Shmelkov (Incremental Learning of Object Detectors without Catastrophic Forgetting) in view of Li (RILOD: Near Real-Time Incremental Learning for Object Detection at the Edge) in further view of Perez-Rua (Incremental Few-Shot Object Detection).
Regarding claim 1, Shmelkov teaches a system comprising: a detection device comprising (figure 2):
classify objects in the image using a dual-task classification model comprising: a pretrained detection branch configured with fixed parameters to classify predetermined objects (abstract and section 3.1); and
a trainable detection branch configured with learned parameters to classify new objects of interest, wherein the trainable detection branch is configured to operate in parallel with the pretrained detection branch (section 3.2 and figure 2);
update the trainable detection branch of the dual-task classification model with the received updated learned parameters while the fixed parameters of the pretrained detection branch remain frozen (section 3.2, frozen A and adapted B).
Li teaches a sensor configured to generate sensor data, wherein the sensor data comprises at least one image and a logic device, the logic device configured to (section 3, system overview);
Li teaches communicate object detection information to a control system in response to a user selection (section 3, system overview);
receive updated learned parameters based at least in part on the communicated object detection information (section 6.4, final model is transferred).
It would have been obvious prior to the effective filing date of the invention to one of ordinary skill in the art to include in Shmelkov the ability to transfer parameters and generate data as taught by Li. The reason is to allow the system to update specific data by user interaction.
Shmelkov and Li do not explicitly teach a plurality of feature extraction layers configured to receive the at least one image and construct feature maps using the at least one image, wherein both detection branches receive the feature maps and classify objects using the feature maps.
Perez-Rua teaches a dual-task classification model comprising: a plurality of feature extraction layers configured to receive the at least one image and construct feature maps using the at least one image (section 3.2, model formulation, the feature extractor takes as input an image and outputs a 3D feature map, and the feature extractor is shared by all the base and novel classes; figures 1-2);
a pretrained detection branch configured with fixed parameters to classify predetermined objects and configured to receive the feature maps and classify predetermined objects using the feature maps (sections 3.2 and 3.2.1, the feature extractor and base-class object locator are trained on the base classes and then fixed, and the base-class locator processes the feature maps at test time to detect the base classes); and
a trainable detection branch configured with learned parameters to classify new objects of interest, wherein the trainable detection branch is configured to receive the feature maps, classify the new objects of interest using the feature maps, and operate in parallel with the pretrained detection branch (section 3.2 and figure 2, novel class codes are learned and convolved with the same feature maps; section 3.1.1, remarks, each class maintains its own prediction heatmap and makes independent detection without interference between old and new classes).
It would have been obvious prior to the effective filing date of the invention to one of ordinary skill in the art to modify the dual-task classification model of Shmelkov and Li such that shared feature extraction layers construct feature maps provided to both the pretrained detection branch and the trainable detection branch as taught by Perez-Rua. The reason is to allow new object classes to be added without interference with or forgetting of the predetermined classes while sharing a single feature extraction computation, which reduces computation and memory usage on the detection device (Perez-Rua, abstract and section 3.2).
Regarding claim 2, see Li, section 1, drones.
Regarding claim 3, see section 1 of Li, camera.
Regarding claims 4-5, see section 3.2 of Shmelkov, class labels and bounding boxes; see also Perez-Rua, section 3.1.1, per-class heatmaps thresholded by activation and decoded into bounding boxes.
Regarding claims 6-8, see Li section 6 which teaches edge devices and updating.
Regarding claims 11-21, see the rejection of claims 1-8.
-
Claim(s) 10 is rejected under 35 U.S.C. 103 as being unpatentable over Shmelkov (Incremental Learning of Object Detectors without Catastrophic Forgetting) in view of Li (RILOD: Near Real-Time Incremental Learning for Object Detection at the Edge) in view of Perez-Rua (Incremental Few-Shot Object Detection) in further view of Zang (20160031559) and Bachrach (20180158197).
Regarding claim 10, Zang teaches an unmanned vehicle adapted to track an object in accordance with user instructions, and determining whether the user has instructed the unmanned vehicle to track the object (abstract and pars. 166, 201, and 226-239, the user selects a target and the UAV automatically tracks the target).
It would have been obvious prior to the effective filing date of the invention to one of ordinary skill in the art to include in Shmelkov, Li, and Perez-Rua the ability to have the user give instruction for a UAV as taught by Zang in order to improve target detection.
Zang does not explicitly teach that the detection device logic is further configured to construct a map based on the at least one image.
Bachrach teaches detection device logic configured to construct a map based on the at least one image (pars. 89-90 and figure 12, sensor data including captured images is processed to generate a 3D map of the surrounding physical environment via simultaneous localization and mapping).
It would have been obvious prior to the effective filing date of the invention to one of ordinary skill in the art to include in Shmelkov, Li, Perez-Rua, and Zang the ability to construct a map from the captured images as taught by Bachrach. The reason is to allow the unmanned vehicle to localize itself and plan trajectories that avoid obstacles while tracking the target (Bachrach, par. 90).
Conclusion
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 nonprovisional extension fee (37 CFR 1.17(a)) 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 mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to HADI AKHAVANNIK whose telephone number is (571)272-8622. The examiner can normally be reached 9 AM - 5 PM Monday to Friday.
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, Henok Shiferaw can be reached at (571) 272-4637. 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.
/HADI AKHAVANNIK/Primary Examiner, Art Unit 2676
Prosecution Insights