Traffic flow prediction system based on video ai analysis

• Main Authors: LI WENWEI, CAO BIRONG, SHAO CHANGCHENG
• Format: Patent
• Language: Chinese; English
• Published: 12.07.2024
• Subjects: CALCULATING; COMPUTING; COUNTING; ELECTRIC COMMUNICATION TECHNIQUE; ELECTRICITY; PHYSICS; PICTORIAL COMMUNICATION, e.g. TELEVISION; SIGNALLING; TRAFFIC CONTROL SYSTEMS

The invention belongs to the technical field of traffic control, and particularly relates to a traffic flow prediction system based on video ai analysis, which comprises a video data acquisition and transmission module, a video data preprocessing module, a target detection and tracking module, a traffic target feature extraction module, a traffic flow prediction module and a result output module. According to the invention, the traffic monitoring video data is acquired in real time and preprocessed, the preprocessed video data is subjected to target detection tracking and motion feature extraction based on the deep learning algorithm, and traffic flow prediction is carried out based on the extracted traffic target motion features in combination with historical traffic data. Moreover, the operation conditions of all monitoring cameras in the corresponding road and the inspection emergency condition can be accurately fed back, and adaptive management schemes are adopted for different road sections in the road,