Notes
Slide Show
Outline
1
Can computers match human perception?
2
Successes
  • Computer vision algorithms rival humans
  • (from class discussion)
    • aligning images
    • face morphing (Conan)
    • super-res, filling in hole, inpainting
    • face recognition with very busy images, volume
    • texture synthesis
    • shape reconstruction?
    • navigation?
    • perception biases, optical illusion
    • pattern matching
3
Challenges
  • CV still far behind
  • (from class discussion)
    • image understanding
    • segmentation
    • 3D shape
    • computers can’t drive?
    • making good photographs (composition)
4
Directions:  Sensors and Imaging
5
Directions:  Detection and Retrieval
6
Directions:  Vision and Learning
7
Directions:  Capturing Humans
8
Directions:  Scene Reconstruction