Stanford CS231A: Computer Vision, From 3D Perception to 3D Reconstruction and beyond

Course Website

• Course Website

Course Overview

CS231A is Stanford’s foundational course on 3D computer vision. It covers the geometric and algorithmic principles behind recovering 3D structure from 2D images. Compared to CS231n’s broad coverage of deep learning for 2D vision tasks, CS231A dives deep into the mathematical and geometric foundations that underlie 3D perception, making it an essential complement for anyone interested in robot vision, AR/VR, or spatial AI.

Also, the latter part of the course introduces cutting-edge topics like neural radiance fields (NeRFs) and Gaussian splatting, which are crucial for modern 3D reconstruction and rendering. While I have a well understanding before for my own research so it may be tedious for me.

My Learning Journey

• 2026.5.14 Lecture1 Introduction
• 2026.5.14 Lecture2 Camera Models
• 2026.6.1 Lecture3 Camera Models II and Camera Calibration
• 2026.7.15 Lecture4 Single View Metrology
• 2026.7.17 Lecture5 Epipolar Geometry
• 2026.7.17 Lecture6 Stereo Systems
• 2026.7.17 Lecture7 Structure from Motion
• 2026.7.19 Lecture8 Active Stereo & Volumetric Stereo
• 2026.7.19 Lecture9 Fitting and Matching
• 2026.7.20 Lecture10 Representations & Representation Learning
• 2026.7.20 Lecture11 Monocular Depth Estimation & Feature Tracking
• 2026.7.20 Lecture12 Learning-based Stereo & Monocular Depth Estimation & Feature Tracking Cont
• 2026.7.21 Lecture13 Optimal and Scene Flow
• 2026.7.23 Lecture14 Optimal Estimation
• 2026.7.27 Lecture15 Optimal Estimation Cont
• 2026.7.27 Lecture16 Neural Radiance Fields
• 2026.7.27 Lecture17 Gaussian Splatting
• 2026.7.27 Lecture18 Guest Lecture