Real Time Automated Sign Language Detection and Translation Using AI and ML Driven Hand Gesture Recognition

This paper discusses an automatically generated real-time framework towards the conversion of Indian Sign Language into text using artificial intelligence and machine learning methodologies related to hand-gesture recognition. This system uses a CNN model to classify gestures and it processes live video input from the webcam to extract hand gesture features using MediaPipe, trying to bridge this gap in communication for the hearing-impaired community with that of the mainstream society. Thorough analyses include the system architecture, methodologies of data collection, model training, and testing processes iteratively, along with challenges such as changing lighting conditions, hand placement, and gesture speed, and strategies used to overcome these challenges. Promising results show high accuracy in real-time recognition, pointing out the potential of the framework to enhance communication accessibility. Subsequent research endeavors will prioritize the augmentation of the system to facilitate word-level recognition, incorporate multilingual functionalities, and improve its robustness for utilization across varied environments, thereby promoting wider applicability and empowering individuals with special abilities.