Self-adaptive correlation semi-supervised learning method based on graph convolution

The invention relates to the technical field of video processing, in particular to an adaptive correlation semi-supervised learning method based on graph convolution, which comprises the following steps: preprocessing a video, and segmenting the video into different video segments according to action categories; extracting features of the video sample data, and dividing the video sample data into labeled data and unlabeled data; constructing a graph structure of the video data based on the features of the video sample data; utilizing the graph structure, introducing a self-adaptive weight calculation method, and obtaining a newest correlation weight according to node feature similarity; learning global topological information from the data with the labels and the data without the labels; performing linear transformation on the source node features, and then aggregating the source node features and adjacent node features to obtain input features; the input features are introduced into the classification layer