Online detection system for electric bike in elevator or corridors based on multi-scale fusion

The traditional elevator building room electric vehicle monitoring system adopts manual monitoring, which has disadvantages such as easy fatigue of personnel and difficulty in monitoring each video channel; to address this situation, we propose an online detection system for elevator building room electric vehicles based on the improved YOLOv4-tiny algorithm, which first crawls pictures of electric vehicles as positive samples, and crawls pictures with relevant backgrounds or similarity in order to reduce the error detection rate and leakage detection rate of the system. The experimental results show that the map of the online detection system based on the improved YOLOv4-tiny algorithm is 94.07%, which is about 2 percentage points higher than the original algorithm.