US 12,260,615 B2
Image attack detection method and apparatus, and image attack detection model training method and apparatus
Bo Li, Guangdong (CN); Jianghe Xu, Guangdong (CN); Shuang Wu, Guangdong (CN); Shouhong Ding, Guangdong (CN); and Jilin Li, Guangdong (CN)
Assigned to TENCENT TECHNOLOGY (SHENZHEN) COMPANY LIMITED, Shenzhen (CN)
Filed by TENCENT TECHNOLOGY (SHENZHEN) COMPANY LIMITED, Guangdong (CN)
Filed on Nov. 30, 2022, as Appl. No. 18/072,272.
Application 18/072,272 is a continuation of application No. PCT/CN2022/086735, filed on Apr. 14, 2022.
Claims priority of application No. 202110431153.1 (CN), filed on Apr. 21, 2021.
Prior Publication US 2023/0104345 A1, Apr. 6, 2023
Int. Cl. G06V 10/764 (2022.01); G06F 21/56 (2013.01); G06V 10/44 (2022.01); G06V 10/74 (2022.01); G06V 10/774 (2022.01); G06V 10/776 (2022.01); G06V 10/80 (2022.01)
CPC G06V 10/764 (2022.01) [G06F 21/566 (2013.01); G06V 10/44 (2022.01); G06V 10/761 (2022.01); G06V 10/774 (2022.01); G06V 10/776 (2022.01); G06V 10/809 (2022.01); G06F 2221/034 (2013.01)] 20 Claims
OG exemplary drawing
 
1. An image attack detection method, performed by a computer device, the image attack detection method comprising:
acquiring an image-to-be-detected, and performing global classification recognition based on the image-to-be-detected to obtain a global classification recognition result;
performing local image extraction randomly based on the image-to-be-detected to obtain a target number of local images, the target number being obtained by calculation according to a defensive rate of a reference image corresponding to the image-to-be-detected, and the defensive rate of the reference image being used for characterizing a defense level of the reference image being attacked by an image;
performing local classification recognition based on the target number of local images respectively to obtain respective local classification recognition results, and fusing the respective local classification recognition results to obtain a target classification recognition result; and
detecting a similarity between the target classification recognition result and the global classification recognition result, and determining the image-to-be-detected as an attack image when the target classification recognition result and the global classification recognition result are dissimilar.