Tracking targets from videos
vectors are generated based on the suspect's overall appearance characteristics, including physical appearance, clothing, accessories and other portrait features, and similar targets are retrieved in video files.
Supports retrieval using uploaded target images or target images brought in from video analysis results. Supported types include images of person, face, vehicles, etc.
Supports fuzzy search of vehicles by entering license plate number
Supports cross-video files retrieval, and the retrieval results can be displayed by video file or sorted by similarity
Support cross-modal retrieval, including walking, non-motorized personnel, squatting and other different types of personnel targets
