Sign Language is an essential form of communication for deaf and hard-of-hearing people, providing a visual interaction system through gestures, facial expressions and body positions. The deep learning- based object detection models have notably improved real-time detection of Indian Sign Language (ISL) signs. This research offers a comparative study of YOLO (You Only Look Once) versions, assessing their performance in terms of accuracy, recall, precision, mean Average Precision at 50% IoU mAP50, and mAP50-95 on a heterogeneous dataset. The results show that the highest performance is obtained by YOLOv8, followed by YOLOv3, YOLOv5, and YOLOv11, with YOLOv8 recording a precision of 0.9765 and an mAP50 score of 0.9751. This study offers practical recommendations on choosing the best YOLO models for ISL recognition tasks and helps narrow the communication gap between hearing and hearing-impaired communities, promoting more inclusivity and accessibility.