四目一体化半球网络摄像机系统设计OA

In order to reduce the complexity of the equipment in the large scene monitoring,and to achieve the continuous tracking of pedestrians in multiple areas,a four-lens integrated dome network camera is designed and developed.The hardware system takes digital signal processor(DSP)as the core,with an eight-megapixel image sensor and a gigabit Ethernet transceiver chip,to realize remote real-time monitoring of large-scene video.In terms of software,multi-camera target tracking algorithm is used to reach multi-area pedestrian continuous tracking task,and the target detection algorithm YOLO(You Only Look Once)is improved in the detection phase.By optimizing the backbone and neck network,and replacing the loss function,the average accuracy of the model is improved by 2.8％,and the computation amount is reduced by about 30％.Through the field scene test,it is verified that the four-lens integrated dome network camera operates stably and reliably,and the average tracking accuracy of the moving target is 91.3％,which can be used for real-time monitoring and pe-destrian tracking in the complex area of large scenes.