Advances in ASL Detection: a YOLOv8-based framework

We present in this paper a YOLOv8-based framework to accurate detection of American Sign Language "ASL". ASL is widely used in bridging the "communication" among hearing communities. The proposed framework provides a real-time detection of gesture in a complex environment. The proposed framework has been development, implemented, and tested using 6033 images with size of 416x416. It contains 36 classes, 26 classes for Letters ('A' - 'Z'), and 10 classes for Numbers ('0' - '9'). The training dataset have been augmented to better increase the accuracy of detection. The simulation results shows that the efficiency of detecting in a high frame rate.