US 12,154,309 B2
Joint training of neural networks using multi-scale hard example mining
Anbang Yao, Beijing (CN); Yun Ren, Beijing (CN); Hao Zhao, Beijing (CN); Tao Kong, Beijing (CN); and Yurong Chen, Beijing (CN)
Assigned to Intel Corporation, Santa Clara, CA (US)
Filed by Intel Corporation, Santa Clara, CA (US)
Filed on Sep. 6, 2023, as Appl. No. 18/462,305.
Application 18/462,305 is a continuation of application No. 17/408,094, filed on Aug. 20, 2021, granted, now 11,790,631.
Application 17/408,094 is a continuation of application No. 16/491,735, granted, now 11,120,314, issued on Sep. 14, 2021, previously published as PCT/CN2017/079683, filed on Apr. 7, 2017.
Prior Publication US 2024/0013506 A1, Jan. 11, 2024
Int. Cl. G06V 10/00 (2022.01); G06F 18/243 (2023.01); G06N 3/04 (2023.01); G06N 3/08 (2023.01); G06V 10/44 (2022.01); G06V 10/82 (2022.01); G06V 20/10 (2022.01); G06V 20/70 (2022.01); G06V 30/19 (2022.01); G06V 30/24 (2022.01)
CPC G06V 10/454 (2022.01) [G06F 18/24317 (2023.01); G06N 3/04 (2013.01); G06N 3/08 (2013.01); G06V 10/82 (2022.01); G06V 20/10 (2022.01); G06V 20/70 (2022.01); G06V 30/19173 (2022.01); G06V 30/2504 (2022.01)] 18 Claims
OG exemplary drawing
 
1. A method for performing object detection, the method comprising:
generating, by executing a machine learning model using at least one processor, respective objectness scores for one or more regions of an image;
selecting a first region of the one or more regions based on an objectness score in response to the first region meeting an objectness threshold;
calculating a localization value for the first region;
calculating a classification score for the first region;
determining a multi-task loss score based on (a) the objectness score, (b) the localization value, and (c) the classification score, the multi-task loss score used to determine whether an object is contained in the first region of the image.