S.I.G.N. - Sign Interpretation and Gesture Navigation
Sania Kalpesh Pimple
Computer Engineering Department
Atharva College of Engineering,
Mumbai University, India
pimplesania-cmpn@atharvacoe.ac.in
Chaitali Parab
Computer Engineering Department
Atharva College of Engineering,
Mumbai University, India
chaitaliparab-cmpn@atharvacoe.ac.in
Devashree Rajesh Karande
Computer Engineering Department
Atharva College of Engineering,
Mumbai University, India
karandedevashree-cmpn@atharvacoe.ac.in
Mann Chirag Shukla
Computer Engineering Department
Atharva College of Engineering,
Mumbai University, India
shuklamann-cmpn@atharvacoe.ac.in
Prof. Nileema Pathak
Computer Engineering Department
Atharva College of Engineering,
Mumbai University, India
nileemapathak@atharvacoe.ac.in
Abstract - The S.I.G.N (Sign Interpretation using Gesture Navigation) project addresses the communication gap between sign language users and non-users, especially in sectors like education, healthcare, and public services.The project places a strong emphasis on Indian Sign Language (ISL) to cater to the communication needs of the hearing-impaired community in India. By tailoring recognition models and sign databases to ISL, the system ensures higher relevance, usability, and cultural alignment for Indian users. It utilizes machine learning, computer vision, and natural language processing (NLP) to translate sign language gestures into meaningful text or speech in real time. The system comprises two key components: (1) Real-time Sign to Sentence Generation, which converts live gestures into accurate, context-aware sentences, and (2) Word to Sign Generation, which displays typed words as corresponding sign gestures for learning and interaction.
Convolutional Neural Networks (CNNs) handle gesture recognition, while MediaPipe and OpenPose ensure precise pose estimation. NLP techniques enhance grammatical accuracy, and the system achieves over 95% accuracy, delivering reliable performance. A Streamlit-based interface enables smooth real-time
interaction, making S.I.G.N a scalable, cost-effective, and inclusive alternative to human interpreters.
Keywords: Computer Vision, Deep Learning, Convolutional Neural Networks (CNNs), Machine Learning, Natural Language Processing (NLP), Image Processing, Feature Extraction, Hand Tracking.
