Video model training method and device and video classification method and device

The invention discloses a video model training method and device and a video classification method and device, and relates to the field of video action recognition. The video classification method and device are invented for solving the problems of high cost and low efficiency of video classification in the prior art. Comprising the following steps: respectively acquiring a plurality of training key frames of each training video in a training video set; respectively obtaining a visible image block and a shielding image block of each training key frame; respectively obtaining code embedding of each training video; respectively embedding the code of each training video into an encoding module in the input video self-supervision model to obtain a first output corresponding to the embedding of the training classification flag bit of each training video, and a second output corresponding to the embedding of the training embedding set and the corresponding position; respectively inputting the decoding flag bit corr