HPMC: A Multi-target Tracking Algorithm for the IoT

With the rapid development of Internet Things and advanced sensors, vision-based monitoring forecasting applications have been widely used. In context Things, visual devices can be regarded as network perception nodes that perform complex tasks, such real-time road traffic flow, target detection, multi-target tracking. We propose High-Performance detection Multi-Correlation measurement algorithm (HPMC) to address problem occlusion trajectory correlation matching for The consists three modules: 1) For module, we proposed You Only Look Once(YOLO)v3_plus model, which is an improvement YOLOv3 model. It has a multi-scale layer repulsion loss function. 2) feature extraction module extracts appearance, movement, shape features. A wide residual model established, coefficient k added extract appearance features target. 3) tracking multi-correlation measures are used fuse extracted increase degree track improve performance. experimental results show method better performance small occluded targets than comparable algorithms.