Kim Anh Phung

Vision-language models can achieve over 96% accuracy in identifying dangerous actions by children, far surpassing traditional deep learning methods.