Gesture Recognition System Based on Improved YOLO v3

In order to realize the normal communication between the hearing impaired and the society, this paper combines the Raspberry Pi and YOLO v3 target detection to achieve the automatic human gesture recognition. The system consists of four parts: image acquisition and transmission module, image preprocessing, training recognition module, speech interaction and front-end display module. In this paper, we combine the image algorithm in OpenCV for image preprocessing, and propose a human gesture recognition system based on improved YOLOV3, to solve the problem of low accuracy and slow speed of the gesture recognition algorithm based on feature extraction in image recognition, and test it in different scenes. The experimental results show that the recognition accuracy of the system with the advanced algorithm is improved compared with that before. The gesture recognition accuracy of the system is around 90%, which means the recognition results are more reliable and can meet the real-time requirements.