A Revolutionizing Deep Sign Language Prediction for Autism Spectrum Disorder Using Hybrid Deep Learning Algorithm and Computer Vision

Sign language is essential for enabling effective communication among individuals who are deaf or hard of hearing. The automatic prediction and recognition of sign language is done by means of machine learning techniques which is essential for improving accessibility and bridging communication gaps. The research proposed in this paper presents a sign language prediction system utilizing a deep learning algorithm SVM, ANN and LSTM-CNN model. Long Short-Term Memory - Convolutional Neural Networks (LSTM-CNN) can be considered one of the best sign language prediction models that excel in processing both spatial and temporal features, making them well-suited for recognizing complex gestures over time. Their effectiveness stems from their ability to integrate motion and hand shape data to accurately predict sign language sequences. This problem statement enhances accessibility by enabling better interaction with technology and people who might not know sign language. It also serves as a technological innovation and promotes inclusivity. This system is designed to identify and classify sign language gestures from video inputs. Key features are extracted from the video frames, including hand shape, position, and motion. These features are then glutted into the LSTM-CNN model for training and classification and implementation accuracy results SVM is 89%, ANN is 93% and high accuracy is predicted in LSTM- CNN is 96%.