Intelligent Television Operating System Based on Gesture Recognition
Huijing Wang, Yimei Xu · 2024
This paper proposes a smart TV control system based on the hardware capabilities of smart TV terminals equipped with neural convolutional network processing power and the transfer learning strategy of Multi Parallel Convolutional Neural Network (MPCNN). The system consists of a gesture recognition module, a semantic determination module, a user interaction interface, and an API interface implementation layer. The gesture recognition module captures user gesture image data through the built-in camera of the TV. The image data undergoes quantization, normalization, and feature extraction processing. The processed data is then compared with the target gesture to give a recognition result. The semantic determination module evaluates the gesture recognition result and the user context to determine the type of user operation. The interaction interface then facilitates interaction between the operating system and the user. Finally, the API interface implementation layer calls system APIs to execute user commands. This technical solution is universally applicable to smart TVs that support the system. It also allows users to customize gestures to expand control functions, achieving system personalization while enhancing fun and improving the user experience.