• Arjit Srivastava
• Aditya Verma
• Abhishek Dubey
• Sadhana Rana

This study presents a real-time sign language recognition system by integrating MediaPipe’s advanced hand tracking with deep learning techniques. We investigate how combining MediaPipe's landmark detection with Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks improves the recognition of both static and dynamic signs. Our findings highlight that a hybrid CNN-LSTM model delivers superior performance compared to individual approaches. Furthermore, we incorporated this system into a mobile application developed with Flutter, utilizing Flask for backend API support and Firebase for user authentication. Experimental results demonstrate an accuracy range of 94–98% for static gestures and 85–92% for dynamic signs, indicating strong potential for applications in education, communication support, and accessibility.

Sign Language Recognition, MediaPipe, Deep Learning, CNN, LSTM, Mobile Application, Flutter, Flask, Firebase.

"Real-Time Sign Language Recognition Using MediaPipe and Deep Learning Approaches: A Mobile Application Integration", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.12, Issue 5, page no.c795-c800, May-2025, Available :http://www.jetir.org/papers/JETIR2505309.pdf

"Real-Time Sign Language Recognition Using MediaPipe and Deep Learning Approaches: A Mobile Application Integration", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.12, Issue 5, page no. ppc795-c800, May-2025, Available at : http://www.jetir.org/papers/JETIR2505309.pdf