US 11,989,924 B2
Image recognition system, image recognition apparatus, image recognition method, and computer readable medium
Shigeaki Namiki, Tokyo (JP)
Assigned to NEC CORPORATION, Tokyo (JP)
Appl. No. 17/609,836
Filed by NEC Corporation, Tokyo (JP)
PCT Filed May 22, 2019, PCT No. PCT/JP2019/020310
§ 371(c)(1), (2) Date Nov. 9, 2021,
PCT Pub. No. WO2020/235059, PCT Pub. Date Nov. 26, 2020.
Prior Publication US 2022/0222916 A1, Jul. 14, 2022
Int. Cl. G06K 9/00 (2022.01); G06T 5/50 (2006.01); G06V 10/72 (2022.01); G06V 10/772 (2022.01)
CPC G06V 10/72 (2022.01) [G06T 5/50 (2013.01); G06V 10/772 (2022.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01)] 8 Claims
OG exemplary drawing
 
1. An image recognition system comprising:
a series image sequence input unit configured to input a series image sequence having sequentiality;
a reference image selection unit configured to select a reference image from the series image sequence;
a proximity image selection unit configured to select a proximity image that is close to the reference image from the series image sequence; and
an inference unit configured to recognize the reference image and the proximity image by performing inference processing on the reference image and the proximity image, the inference processing comprising convolution processing and activation function processing for outputting a constant value with respect to an input value within a predetermined region, wherein
the inference unit generates results of performing the convolution processing and the activation function processing on the proximity image from the results of the convolution processing and the activation function processing performed on the reference image, and the result of the product of the convolution processing performed on a difference image, which is an image of the difference between the reference image and the proximity image, and a derivative value of the results of the convolution processing and the activation function processing performed on the reference image.