Computer Vision

Computer Vision

1. Challenges

   • Deformation
   • Illumination
   • Occlusion
   • Scale
   • Viewpoint
   • Clutter
   • Motion
   • Intra-Class Variation
   • Local Ambiguity

2. Cues

   • Depth
     • Linear Perspective
     • Aerial Perspective
   • Shape and lighting
     • Shading
   • Grouping
     • Common Fate

3. Connected to:

   • ML
   • AI
   • Robotics
   • Image Processing

4. Vision

   • I. Basic Processing - feature detection
   • 2 Fitting & Grouping
   • 3 Multi-view Geometry
   • 4 Recognition
   • 5 Semantic Segmentation
   • Object Recognition
     • Problems
       • Post Est. Where is it in 3d?
       • Segmentation - Where is it in 2d
       • Recognition - What is it
     • Attempts
       • Recognition from appearance
       • Recognition from Geometry
       • Best: view based, local invariance,' lots of features and voting
     • Features
       • Distinctive
       • many per object
       • Easy to detect
   • Object Categorization
     • MAP - bayes
   • Face Detection
     • Slide window across image - evaluate face model

5. Robotics

   • Where am i
     • SLAM
   • Whats nearby
     • Navigation/Learning

6. Perception

   • Computation STereo
   • Feature Detection
   • Motion Tracking