An End-to-End Sign Language Translation Pipeline from Static Gestures to English Using T5

Sign language remains a crucial mode of communication for individuals with hearing and speech impairments, yet it remains largely inaccessible to the general population. This paper addresses the problem of translating finger spelled American Sign Language (ASL) from static images into fluent English sentences. We propose an end-to-end pipeline that integrates gesture recognition with natural language processing (NLP). Hand gestures are classified using Google Teachable Machine and refined with MediaPipe for better hand tracking and landmark detection. The classified signs are then passed to a fine-tuned FLAN-T5-small model to generate grammatically correct English sentences. This paper demonstrates a functional prototype that successfully bridges gesture inputs with natural language generation. Evaluation using BLEU and ROUGE-L scores indicates that the system produces coherent and contextually relevant translations, showcasing its potential as an assistive communication tool.