Control Home Appliances Using Hand Gesture Recognition Based on Deep Learning and Arduino Uno via Infrared Signal
Year: 2025
Type: article
Abstract: Currently, deep learning (DL) has been applied to various fields of reality and yielded very convincing results, especially in problems related to computer vision and natural language processing. Controlling devices with AI technologies is not a new problem, but there are still challenges in terms of accuracy, latency, and application in conditions of low-configuration, low-cost devices, and low-quality input data. In this paper, we conduct research on proposing a solution to build a system to control devices in smart homes (Samsung TV, the Daikin air conditioner, and the Vinawind fan) using hand gesture recognition (HGR). Our proposed system includes the step of building control commands (6 gestures corresponding to 6 control commands) and connecting, sending commands to execute control commands using Arduino UNO via infrared signals to the device. The process of creating commands is based on using YOLOv11 for HRG. Choosing YOLOv11 is based on comparative research on HGR results with the methods: Mediapipe, YOLOv11, YOLOv12, MobileNetV3-Large, Transformer-Based Network (TBN), Multiscale Video Transformer Network (MVTN), and GestFormer. YOLOv11 has near-perfect accuracy (acc=99.5%, P=99.3%, R=99.9%, and F1 − score=99.6%), and YOLOv11 is capable of working well with low-configuration computers, and calculations are performed on the CPU. The delays in device control for Samsung TV, Daikin air conditioner, and Vinawind fan are 0.14s, 0.18s, and 0.18s, respectively. Our proposed system is capable of operating according to the requirements of reality. The results and demos of the system are readily available.
Institution: Trường Đại học Tân Trào
Language: English
Subfield: Human-Computer Interaction
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W7160922262
