← All Days
Day 1 — Sat, Mar 21
How does the brain begin processing vision?
- Goal: understand the human visual cortex — the brain is the original inspiration for computer vision
- Image processing in the brain begins at the Primary Visual Cortex, which splits into two streams
- Dorsal stream (the "where" pathway) — answers "where is this thing?" and "what do I do about it?" — extremely fast
- Ventral stream (the "what" pathway) — identifies objects via color, texture, shape — takes longer to process
- The retina has two types of output cells: M-cells (fast) and P-cells (slow) — this is why one stream is faster than the other
- Dorsal stream works with M-cells, which are designed to detect changes in the scene — it starts at the retina itself
- MT area (Middle Temporal) is where motion is tracked and changes in the image are processed
- PPC (Posterior Parietal Cortex) is the GPS of the brain — builds a spatial map of the world: distance, speed, relative position
- PPC splits its output into action (reaching, grasping, dodging) and gaze (where to move eyes next, via motor/pre-motor cortex)
- Dorsal stream operates in real-time and has no memory — picking up a pen triggers automatic computation of pressure & depth before conscious awareness