Introduction

1. 什么是 Vision? Sensation - Processing - Perception - Cognition, 包括神经/识别的 Perception 是更重要的部分
2. Visuomotor coordination
3. Vision 是 Modality 中较重要的一部分, 人有 83% 的感知来自 vision
4. 人的 Perception-action(body control) Loop 是闭环的, vision 指导 action, action 反馈 vision, 进而改正, 这是人与动物都有的行为 具身智能
5. Vision 与 language 交互(视觉语言/文字)

Computer Vision 做人类做的很多事情, 做人类不能做的很多事情

Computer Vistion 不是 Deep Learning 的应用, 是人本的, 是从神经科学出发的

Technology & History

What is Computer Vision? It deals with:

• acquiring
  • RGB camera, Depth camera and LiDAR (maybe others)
  • Stereo, even multiview Images, Panoramic images
  • Video
  • True 3D: Point Cloud, mesh, and volumetric
• processing and analyzing: Low-level
  • Image processing
    • Applications...
  • Future extraction
    • edge/corner cognition
• Mid-level Vision
  • grouping
  • 3D reconstruction
  • Scanner
  • SLAM
• High-level Vision: Understanding, semantic interpretations
  • Facial Recognition
  • Scene Understanding
  • Augmented Reality
• Graphics and Vision - 前沿
  • 图形学的 Synthetic data (differ from Real data) 与 Vision 对抗?
  • NeRF
  • Vision 还要管人不管的事情: Generation

  What I cannot create, I do not understand. - Feynman

  • Animating Faces
  • Diffusion Model
  • Vision-Language 待发展
• Embodied AI
  • Autonomous Driving 端到端前些年都是for PR purposes Entering l3
  • Generalist AGI
  • (He Wang) GraspVLA