The system uses monocular camera to collect images, uses face detection to determine the target position, uses the face position of the current frame image to determine the air handwriting area and combines hardware PTZ control to achieve target tracking, and then uses image recognition technology to achieve air gesture segmentation, air gesture recognition, air handwriting track collection, and air handwriting track recognition for the air handwriting area image, finally, according to the air gesture, handwriting recognition results to achieve multimedia teaching PPT page turning and character input and other intelligent teaching operation. The main work of this system is as follows: (1) Through the image acquisition module to obtain the image, in the target tracking module, the use of Harr-like features, Adaboost algorithms and other methods to achieve a faster face detection, according to the face detection results to determine the air handwritten area and steering gear rotation angle. The hardware control module is built by STM32, PCA9685 and steering gear to assist in the realization of face target tracking in intelligent teaching. (2) In the air gesture recognition module, the air handwriting area image is segmented by using the inter-frame difference and GMM model, and the air gesture recognition model is built based on the PCA-SVM. The experimental results show that the air gesture recognition rate is 93.44, and the average time per frame is 0.04 seconds. (3) In the air handwritten track acquisition module, a new fingertip detection algorithm is proposed for air handwritten fingertip detection. The experimental results show that the performance and recognition rate of the same test data are better than the traditional center of gravity distance method and other fingertip detection, and the recognition rate is 95.81%, and the average time per frame is 0.0009 seconds. (4) In the air handwriting recognition module, an air handwriting recognition model based on 3-layer convolutional neural network is designed and implemented, and the Top-4 method is adopted to increase the recognition rate from 94.64 to 98.79, and the average time of writing a character in the air is about 2 seconds.