Real-Time American Sign Language Detection Using YOLOv5 and YOLOv8: Analytical Comparison

Culture and religion today are many and spread around the world. Since it began in 1817 at the American School for the Deaf (ASD), sign language culture has grown. Deep learning is used in computers today to solve problems and make apps that work in real time. Sign language (SL) is one of these. Convolutional neural network (CNN) is used in YOLO, an object detection and classification algorithm, to make it work quickly and correctly. The paper uses a custom model for sign language recognition to try to find American sign language using YOLO models and compare different YOLO algorithms. The tests showed that the new YOLOv8 did better than other versions of YOLO in terms of accuracy and mAP. However, YOLOv7 did better on the recall test than YOLOv8. The suggested model is quick and doesn't take up much space. It tests and trains on the American Sign Language letters dataset. It was able to achieve a 97 percent recall rate, a 95 percent precision rate, and a 96 percent mAP @0.5, which means it can recognise hand gestures in real time.