YOLO-V4 Based Detection of Varied Hand Gestures in Heterogeneous Settings

The detection of varied hand gestures in heterogeneous settings plays a crucial role in enhancing human-computer interaction. Leveraging the YOLO-V4 model for object detection offers the potential for faster performance and improved accuracy, which are paramount in the field of Artificial Intelligence. This paper presents an approach to training custom datasets with YOLO-V4, enabling efficient detection of hand gestures with minimal data requirements. Unlike traditional methods that rely on large datasets, our proposed model achieves faster training times and higher accuracy levels, addressing the limitations of prolonged training durations and reduced accuracy. By focusing on eloquent and commonly used Indian subcontinental hand gestures, encompassing single and dual-hand gestures, we aim to enhance man-machine communication. Ten distinct datasets are meticulously labeled, trained, validated, and tested using the YOLO-V4 model. Comparative analyses against existing models further validate the efficacy and superiority of our approach. This research contributes to advancing the field of object detection by demonstrating the efficacy of YOLO-V4 in detecting hand gestures with limited datasets, thereby facilitating more efficient and accurate human-computer interaction.