Real-Time Hand Gesture Recognition and Sign Language Detection

Abstract: This paper presents a real‑time system for recognizing American Sign Language gestures using deep learning. A webcam captures hand gestures, which are classified into 26 alphabets (A–Z) and three symbols (DEL, SPACE, NOTHING). The system employs a convolutional neural network with transfer learning from a pre‑trained ImageNet model, eliminating manual feature extraction. Implemented in Python using TensorFlow and OpenCV, the pipeline preprocesses live frames and performs instant gesture prediction. Experimental results demonstrate high classification accuracy and smooth real‑time performance under normal lighting conditions. This affordable, hardware‑free solution serves as an assistive tool to improve communication between hearing‑impaired individuals and non‑signers.

Keywords: American Sign Language recognition, Convolutional Neural Networks, transfer learning, real-time gesture classification, assistive communication technology, computer vision