In our daily life, in addition to using language and words to communicate, we often use gestures to convey information. For example, we will wave to say hello or farewell, thumbs up to express praise, traffic police will also use gestures at the intersection to guide the traffic. Therefore, gestures have become an important means of communication between people. In recent years, the related algorithm of gesture interaction has been widely used in various AR glasses, and its ultimate goal is to integrate it into the system layer of AR glasses to replace the handle to complete most of the interaction requirements. If we can free gesture recognition from the limitations of AR hardware and apply it to related devices in daily life, it may bring great convenience to our lives. For example, when we watch TV, if we want to switch channels or adjust the volume, we can directly use gestures to operate. When driving a car, if you want to change the music, the driver usually needs to look away from the road, which can be potentially dangerous, and the use of gesture recognition can interact with the car audio without contact. In the smart home environment, we can even use gestures to control lights, air conditioners and other devices. In recent years, due to the outbreak of the new crown epidemic, the health problems caused by contact operation have become more prominent, so people need a non-contact human-computer interaction system. Therefore, we propose a solution called "deep learning-based intelligent gesture control system. "Intelligent gesture control system based on deep learning" is a technology that controls devices through gestures, realizing "interactive communication" between devices and users ". Users can directly access and use the system through the browser, to achieve cross-platform convenience, can be used anywhere, any time. The system captures the user's gestures and hand movements through the camera, and then performs deep learning analysis in the background to quickly respond, replacing the hardware device represented by the mouse, and realizes the movement, click, zoom-in, zoom-out and other operations of the control object. This operation is simple and the development cost is low. "Intelligent gesture control system based on deep learning" can be applied to consumer electronics, government departments, finance and banking, education, medical and automotive industries. It will enhance the digital level of various industries and promote the development of gesture recognition and contactless sensing market.