Implementation of Mask-wearing Detection for Embedded Platform

With the vigorous development of target detection technology, many mask-wearing detection algorithms have come into being, but the detection task still has the problem of being difficult to detect at the embedded platform. In this regard, a mask-wear detection model based on improved YOLOv5 is proposed, and the implementation of the embedded platform of detection was studied. Firstly, the implementation of the mask-wearing embedding terminal was analyzed; then the mainstream embedded platform was compared, and the NVIDIA Jetson Nano was selected for the embedding terminal deployment; finally, the tensorRT acceleration principle was introduced and the acceleration experiment was completed. The results show that the mAP of the post-feed algorithm reaches 99.6% and the FPS is 30, and the demand for mask-wear detection can be met on the embedded platform NVIDIA Jetson Nano, which has reference significance for practical applications.