TY - JOUR AU - T, Kujani. AU - Kumar.V, Dhilip PY - 2022/10/01 Y2 - 2024/03/29 TI - Multiple Deep CNN models for Indian Sign Language translation for Person with Verbal Impairment JF - International Journal of Intelligent Systems and Applications in Engineering JA - Int J Intell Syst Appl Eng VL - 10 IS - 3 SE - Research Article DO - UR - https://ijisae.org/index.php/IJISAE/article/view/2179 SP - 382–389 AB - <p>Understanding the sign language is very useful for people with verbal and hearing impairment. Sign language is a category of nonverbal communication for people weakened by speech and listening capability. Automatic translation of sign gestures into text have gained more attention in recent years. In this research work, a deep CNN based approach has been proposed for detecting existing 35 signs of Indian Sign Language (ISL) alphabets into text in an efficient manner using hand kinematics. A Convolution Neural Network (CNN) with custom developed number of convolution layers with a suitable optimizer is applied. CNN handle issues of lighting conditions and most importantly it is efficient in tackling the problems under computer vision. It is considered for detecting features with required training without any manual pre-processing. The proposed approach has achieved 92.85% of accuracy with ISL dataset and accuracy of this method is compared with the transfer learning models such as Densenet201 and Resnet50.</p> ER -