Video data structuring method based on semantic features and related device

The invention discloses a video data structuring method based on semantic features and a related device, and the method comprises the steps: carrying out the fixed-time frame-interval extraction operation of a target video, and obtaining a plurality of frames of to-be-processed images; performing feature extraction on a target object and a background in the to-be-processed image by adopting a preset AI model to obtain object feature information and environment feature information; performing multi-feature fusion according to the object feature information and the environment feature information to obtain target fusion feature information; semantic conversion processing is conducted according to the target fusion feature information through a preset recognition model, semantic feature information is obtained, and the preset recognition model comprises target detection, posture detection, action recognition and time sequence action detection; and performing structured processing based on the semantic feature in