Pedestrian multi-target tracking algorithm based on multi-feature fusion enhancement

The invention discloses a pedestrian multi-target tracking algorithm based on multi-feature fusion enhancement. According to the algorithm, a feature extraction module, a target detection module and a data association module are integrated into a network architecture. According to the algorithm, two continuous frames serve as a chain node to serve as input, in order to reinforce a resnet50 backbone network, traditional convolution is replaced with Inception convolution, and the receptive field of a feature extraction network is enhanced; in order to enhance the adaptability to target deformation, the extracted features are subjected to weighted bidirectional pyramid fusion processing, and more attention is paid to important feature information; in order to enable the network to better process pedestrian targets in a crowded scene, a context-sensitive prediction module is added into the network, and wider and deeper convolution prediction modules are placed at the top of a layer with different synchronization