Publication

Abstract : The usage of Vision Transformer (ViT) for drowsiness detection is being shown by our work, which uses its uncommon attention mechanisms to detemine complex spatial correlations in facial images. Our method achieves a high accuracy of 99.8%, surpassing traditional approaches similar to MobileNet V2. Our work utilized Driver Drrowsiness Detection(DDD) Dataset. The ViT's excellent performance is associated with its ability to learn detailed and complex patterns from facial features, which highlights its effectiveness in picture classification tasks and its ability to handle real-world issues related to driver safety. This work highlights the adaptability and power of ViT in enhancing the precision and trustworthiness of sleepiness detection or recognition systems, giving more opportunities for further developments in the future.