This internship research project aims to explore cutting-edge deep learning methods in various fields, including autonomous vehicles, video processing, computer vision, VLM, and multimedia systems. Specifically, it covers topics such as small object detection for ADAS systems, focusing on precise identification of small objects in autonomous driving environments; LiDAR-based and Radar-based 3D object detection and multi-sensor fusion (incorporating LiDAR, cameras, and radar) to enhance 3D scene understanding; signal enhancement and recovery for images and point clouds to improve imaging accuracy; 3D scene representation and novel view synthesis to enable realistic scene reconstruction and view transformation; exploration of mixed reality (MR) and virtual reality (VR) content generation to support intelligent buildings and multimedia systems; and integrating generative AI and visual language models to enhance the functionality of surveillance systems and multimodal generation applications. These research topics aim to fully leverage deep learning's potential in multimedia and visual scene processing, with applications extending from autonomous driving to medical image processing, intelligent environments, and emerging VR application scenarios.