Hand Gestures of American Sign Language to English Letters Conversion

Soumyadeep Das, Kalyani Mali

Abstract: American Sign Language (ASL) recognition through computer vision has emerged as an important area in assistive technologies. This research focuses on converting American Sign Language (ASL) hand gestures into corresponding English alphabets using a CNN-LSTM-based architecture. The system leverages MediaPipe's real-time hand tracking capabilities to extract hand landmarks and applies a CNN classifier for accurate gesture classification Experimental evaluations are conducted on the ASL Alphabet dataset.

Keywords: ASL, Deep Learning, CNN, LSTM, Gesture Recognition, Sign Language Translation