Foundations of CV

This project builds a set of supplemental tutorial sections for Foundations of Computer Vision by Antonio Torralba, Phillip Isola, and William T. Freeman. The book explains almost every idea with a figure or a plot. Your job is to reproduce those figures as runnable code, so each concept becomes something you can execute, change, and inspect for yourself.

​

The task

Pick a chapter from the list below and write a tutorial section that reproduces its key figures. Treat the book as the specification: for every plot the chapter uses to make a point, write the code that regenerates it from scratch and explain the concept the figure illustrates. The result is a companion page that sits next to the book and moves a reader from “read the figure” to “run the figure”.

​

What a finished section looks like

• It reproduces the chapter’s main figures with runnable code.
• It states the concept each figure demonstrates and the math behind it.
• It runs end to end, so you can change a parameter and watch the plot update.

Follow the conventions already used across this site:

• One chapter per section, each in its own directory. Reserve that directory’s images/ folder for block diagram artifacts only: the diagram source (Mermaid or C4) together with its rendered .png. Notebook output figures (the plots your code generates) are embedded in the page directly and do not need to be stored under images/.
• Markdown cells and inline comments never discuss the plotting library or the toolchain (matplotlib, headless rendering, install steps, and the like). See the coding guidelines below for what they should emphasize.
• Pure-plotting cells are tagged so the page shows the figure but hides the drawing code.
• The prose addresses you directly and stays evergreen.

​

Coding guidelines

• Write all computational code in PyTorch and libraries built on the PyTorch foundation, such as Kornia for differentiable computer vision. Express the math each figure illustrates with tensors and these libraries rather than NumPy or framework-specific equivalents, so every section runs the same way on CPU or GPU.
• Markdown cells and inline comments explain the code first and the concept second: describe what each block does, then connect it back to the idea the figure illustrates.

​

Where your work goes

Submit your work to the pantelis/eng-ai-agents repository through a fork and pull request.

1. Fork pantelis/eng-ai-agents to your own account and clone your fork.
2. Branch. Create one branch per section, named mit-book-chapter-<chapter>-<section>. For chapter 12, section 5, the branch is mit-book-chapter-12-5.
3. Location. Place every notebook under the notebooks/ folder.
4. Pull request. Push the branch to your fork and open a pull request against main of pantelis/eng-ai-agents.

​

Chapters to treat

The chapters below are grouped by the parts of the book. There are 28 chapters in scope. The Assignee column records who has claimed each chapter; all 28 are currently claimed. Any chapter shown as Open is still available, so add your name in the course channel to take it.

​

Image formation

​

Foundations of learning

​

Image processing

​

Sampling and multiscale image representations

​

Neural architectures for vision

​

Generative image models and representation learning

​

Understanding geometry

​

Understanding motion

​

Understanding vision with language

​

Reference

Foundations of Computer Vision, Antonio Torralba, Phillip Isola, and William T. Freeman, MIT Press. Read it online at visionbook.mit.edu.

---