提取二维卷积特征，需要先对视频进行分帧，然后再对视频帧进行二维卷积(To extract two-dimensional convolution features, we need to segment the video first, and then two-dimensional convolution of the video frame)