Computer Vision Homeworks

Python repository for Computer Vision coursework assignments.

Setup

Requires uv.

uv sync

This creates .venv/ and installs all dependencies.

Running Q1 Data-Loading Scaffold

uv run python scripts/hw01/run_q1.py

Expected output: a summary of loaded shapes and paths. The reconstruction algorithm is not implemented yet — that is intentional.

Lint and Tests

uv run ruff check .
uv run pytest -q

Repository Layout

data/hwXX/          — raw assignment data (never overwrite)
notebooks/hwXX/     — submission-oriented Jupyter notebooks
src/cv_homeworks/   — importable Python package
  hwXX/             — assignment-specific modules
scripts/hwXX/       — runnable scripts for each question
outputs/hwXX/       — generated outputs (gitignored, .gitkeep placeholders)
tests/hwXX/         — pytest tests
docs/hwXX/          — implementation plans and notes

Raw Data

Path	Contents
data/hw01/q1_house/	House dataset: two views, camera matrices, matched points, reference outputs
data/hw01/q2_vase/	Vase GIF for Q2

File-type note: house_1.jpg and house_2.jpg are PNG images despite the .jpg extension. The raw filenames are preserved. Image loaders use Pillow (content-based decoding) so the mismatch is transparent.

Reference outputs (matches_connected.jpg, matches_xy_projected.jpg, reconstruction.gif) live in data/hw01/q1_house/ and must not be overwritten. Generated outputs go to outputs/hw01/q1/.

Notebooks

Path	Status
notebooks/hw01/q1_metric_reconstruction.ipynb	Template — data loading ready, algorithm TODO
notebooks/hw01/q2_sfm_novel_view_synthesis.ipynb	Placeholder

Generated Outputs

Generated files go under outputs/ (gitignored). Only .gitkeep placeholders are tracked so the directory structure exists in a fresh clone.

Current Status

Component	Status
Repository scaffold	Done
uv project + deps	Done
Path constants	Done
Q1 data-loading infrastructure	Done
Q1 algorithm (triangulation, GIF, etc.)	Not implemented
Q2	Not implemented

PyCharm Setup

1. Open the repository root in PyCharm.
2. Run uv sync in a terminal.
3. Select the interpreter at .venv/bin/python (Settings → Project → Python Interpreter → Add → Existing → .venv/bin/python).
4. Mark src/ as a Sources Root if PyCharm does not infer it automatically (right-click src/ → Mark Directory as → Sources Root).
5. Run tests with uv run pytest -q.
6. Run lint with uv run ruff check ..
7. Run scripts from the repository root, for example:

   uv run python scripts/hw01/run_q1.py