Hand Gesture Recognition System Using Thermal Images: A Systematic Literature Review

Leading the way in human-computer interaction (HCI) advancements are hand gesture recognition (HGR) systems, which provide touchless, intuitive control for a wide range of applications, including healthcare, smart devices, assistive technologies, and virtual environments. In order to demonstrate the development of HGR methodologies, such as vision-based, skeleton-based, and hybrid approaches, this paper summarizes the results of recent studies. The field is dominated by vision-based systems, which use depth sensors and RGB cameras to recognize both static and dynamic gestures. Particularly for continuous gesture recognition, skeleton-based methods such as the TD-Net architecture offer computationally effective and lightweight solutions. While multimodal systems that combine RGB, thermal, and depth data improve robustness, thermal imaging expands the applicability of HGR to low-light conditions. Despite advancements, problems still exist, such as signer-independent performance, environmental adaptability, and gesture spotting in continuous streams. Directions for the future. Keywords— Machine learning, Deep learning, Hand Gesture, image processing, Convolutional neural networks, Thermal images.