Introduction to Computer Vision (DATA)
data science
about this course
How does your phone straighten a photograph of a document? How do panorama apps stitch
multiple images seamlessly? How do self-driving cars understand depth from stereo cameras? This
course bridges the mathematical foundations from Geometry for Computer Vision I to real-world
computer vision applications.
Building on your solid understanding of projective geometry from Course I, you'll learn how cameras
capture 3D scenes as 2D images and how to reverse this process computationally. Through hands-on
programming projects, you'll implement the algorithms behind everyday vision applications, from
document scanning to 3D reconstruction from photos.
The course balances mathematical understanding with practical implementation. You'll derive key
algorithms to understand the mathematics, then implement them in Python to see how they work
on real data. A substantial optimization primer ensures you understand the mathematical
foundations underlying computer vision algorithms, what is a cost function, how gradient descent
works, why RANSAC handles outliers.
No advanced mathematics required, just the geometry from DATA239/COMP239, linear algebra,
basic calculus, and willingness to learn Python. Through weekly programming assignments, you'll
build proficiency so that by course end, you'll have implemented core algorithms from scratch and
built a substantial team project suitable for your portfolio.
This course sets the foundation for understanding both classical geometric computer vision and
modern learning-based approaches. Whether you're interested in AI, robotics, AR/VR, or just want
to understand the technology you use daily, this course offers a unique blend of rigorous
mathematics and hands-on application.
in these programmes
- computer science
elective · 4-year major, 4-year double major, 3-year major, 3-year double major, 4-year minor, 3-year minor
planning & teaching
recently taught by
- Shankar SubramanianVisiting Professor, Data Science
Recorded in 2025 T3. Instructors may change.