Computer Vision, Efficient Video Understanding, Multimodal Large Language Model