Application of Gesture Recognition Technology Based on Deep Learning in Intelligent Laboratory

Gesture recognition technology has great value in real life applications, and it is still not mature and has many difficulties to overcome. Therefore, gesture recognition technology still has high research value. The core of this paper is to design a feasible intelligent laboratory, and improve the image segmentation and gesture tracking algorithms to improve the accuracy of gesture recognition. SSD network replaces all the connected layers of the front-end basic network with convolution layers, and extracts the result of one convolution layer for target detection. In this paper, the common convolution layer is split, and each convolution group is divided into a set of deep convolution and a set of point convolution, and each layer has its own convolution kernel by deep convolution operation. The results show that the test accuracy of the proposed algorithm is 97.406%, and it shows better classification and recognition ability when running on the self-built database. DL (Deep learning) will automatically learn the characteristics of gestures in the training set, and when the whole task is implemented end-to-end, regression model will be used to solve the detection task, and the detection result will be significantly improved. Can realize certain market application.