Video stream marking method, device and system based on SGAN and storage medium

The invention discloses a video stream marking method based on an SGAN, and the method comprises the steps: obtaining a new video stream after the training of an SGAN model is detected to be completed, and carrying out the extraction of the new video stream, and obtaining a corresponding target frame picture; and inputting the target frame picture into the SGAN model to obtain a classification label corresponding to the target frame picture so as to complete a marking process of the target frame picture. The invention further discloses a video stream marking device and system based on the SGAN and a storage medium. According to the invention, the trained SGAN model is utilized to classify the frame pictures in the video stream, and the characteristics of the SGAN model are fully utilized, so that the workload of marking training data is greatly reduced. 本发明公开了一种基于SGAN的视频流标记方法，包括：在检测到SGAN模型训练完成后，获取新视频流，对所述新视频流进行提取处理，得到对应的目标帧图片；将所述目标帧图片输入所述SGAN模型，得到所述目标帧图片对应的分类标签，以完成所述目标帧图片的标记过程。本发明还公开了一种基于SGAN的视频流标记装置、系统及存储介质。