CS 180 Computer Vision
Published:
Archive
Notes and Cheat Sheet
Review
A class with particularly interesting content. We learned about cameras, image processing (viewing an image as a 2D signal), homographies, RANSAC, stereo and SfM, NeRFs, and diffusion. No homeworks in the class - instead, projects with minimal starter code. Instruction was average, the discussions were brand new and deteriorated in quality when the neural unit started. Prof. Efros and Kanazawa are definitely some of the GOATs their field. Efros especially has strong opinions on the importance of data, and spent half a lecture glazing Nearest Neighbors.
Overall, CS 180 is a 3.5/5 in difficulty (4.5 for the NeRF and diffusion units if you don’t have significant ML experience). Definitely 4.5/5 in fun. Unfortunately, I choked on the last 2 projects, but did well enough on the exams to end with an okay grade. I’ll take the graduate follow up in the spring so will definitely need to review the units I did poorly on.
