Kranti Kumar Appari’s team integrates a Convolutional Neural Network with computer vision techniques to detect hand landmarks from webcam input, translating British and American Sign Language into readable text or speech. They train on hybrid datasets and apply dynamic preprocessing to handle lighting and backgrounds, ensuring reliable real-time performance for inclusive communication platforms targeting users with hearing impairments.

Key points

• Integration of CNN models with computer vision for real-time detection of sign language gestures, using backpropagation for model optimization.
• Implementation of dynamic preprocessing (lighting normalization, background removal) to ensure robustness across diverse environments.
• Hybrid training dataset combining public sign language repositories with custom gesture images for both British and American Sign Language, enhancing linguistic versatility.

Why it matters: Real-time AI-driven sign language detection democratizes communication access for the hearing-impaired, enabling seamless interaction without the need for manual interpretation.