A Sign Language Recognition System Using Artificial Intelligence

Indian Sign Language (ISL) is a visual language used by millions of people in India to communicate. However, not everyone knows how to interpret it, which can lead to communication barriers and difficulties in various settings such as schools, hospitals, and public institutions. Therefore, we have developed a real-time method using neural networks for finger spelling based on Indian Sign Language. The developed system uses computer vision and deep learning techniques to recognize ISL gestures in real time and convert them into text to facilitate communication. This project has been developed using Yolov5, using the YOLOv5 algorithm for object detection and recognition. OpenCV, an open-source computer vision library, is used for pre-processing and image-processing tasks such as background subtraction and hand segmentation. The custom dataset used to train the model includes a wide range of ISL gestures, including letters of the alphabet and commonly used words and phrases. The model is trained using transfer learning, in which a pre-trained model is fine-tuned on the user-defined dataset. The resulting model achieves over 95% accuracy in the test set. The developed system can recognize and render ISL gestures in real time via a webcam or other video input device.