VIDEO CONTENT RECOGNITION METHOD AND APPARATUS, STORAGE MEDIUM, AND COMPUTER DEVICE

摘要

A video content recognition method comprises: acquiring a video frame set from a target video, and extracting an image feature corresponding to the video frame set; dividing, on the basis of multiple channels of the image feature, the image feature into multiple image sub-features; determining from the multiple image sub-features, on the basis of a preset sequence, image sub-features to be processed; fusing a current image sub-feature to be processed with a convolution processing result of a previous image sub-feature, and performing convolution processing on a fused image feature, so as to obtain, for each image sub-feature to be processed, a corresponding convolution-processed image feature; stitching, on the basis of multiple channels of the image features that have undergone convolution processing, the multiple image features that have undergone convolution processing, so as to obtain a stitched image feature; and determining, on the basis of the stitched image feature, video content corresponding to the target video.