Introduction to Computer Vision – Homeworks (Innopolis University, Fall 2025)

This repository contains my solutions to three homeworks from the “Introduction to Computer Vision” undergraduate course at Innopolis University (Fall 2025). The course gives a practical introduction to core computer vision concepts and their implementation in Python, with a focus on applying CV techniques to real-world problems.

⚠️ Academic honesty / usage notice
All notebooks and code here are my personal coursework solutions. They are provided only for reference and educational purposes.
If you are taking this or a similar course, do not copy these solutions or submit them as your own work.

⚠️ Дисклеймер (по-русски)
Материалы репозитория предназначены исключительно для ознакомления и в качестве справочного примера.
Пожалуйста, не используйте этот код как готовое решение учебных заданий и не сдавайте его от своего имени.

---

Repository structure

.
├── hw1
│   └── TUZOV_ARTYOM_HW1.ipynb
├── hw2
│   └── TUZOV_ARTYOM_HW2.ipynb
├── hw3
│   ├── data/
│   │   ├── HumanL.jpg / HumanR.jpg      # stereo pair with a person
│   │   ├── L5.jpg ...  L25.jpg          # left camera images
│   │   └── R5.jpg ...  R25.jpg          # right camera images
│   ├── data.zip                         # archived copy of the data folder
│   └── TUZOV_ARTYOM_HW3.ipynb
├── LICENSE
└── README.md

All homeworks are implemented as Jupyter notebooks (.ipynb) and use standard Python CV stack (NumPy, OpenCV, Matplotlib).

---

Homework 1 – Image Filtering & Binary Vision (hw1/)

Focus: basic image processing, filtering, and binary vision.

Main ideas and tasks:

• Loading and visualizing images; converting between color and grayscale.

• Implementing 2D convolution “from scratch” and comparing with OpenCV.

• Applying linear filters:

  • box / mean filter,
  • Gaussian smoothing,
  • basic sharpening / unsharp masking.

• Edge detection:

  • gradient computation,
  • Sobel / Laplacian filters,
  • visualization of gradient magnitude and orientation.

• Binary vision:

  • intensity thresholding (fixed, Otsu, etc.),
  • simple morphological operations to clean up binary masks (erosion, dilation, opening/closing).

All code for this homework is in hw1/TUZOV_ARTYOM_HW1.ipynb.

---

Homework 2 – Local Features & Hough Transform (hw2/)

Focus: detecting local features and using the Hough transform for geometric structures.

Main ideas and tasks:

• Keypoint detection with classical feature detectors (e.g., Harris / Shi–Tomasi, ORB/SIFT-like features via OpenCV).

• Building local descriptors and matching keypoints between two images.

• Estimating simple geometric transformations (e.g., homography) from matches and visualizing inliers/outliers.

• Using the Hough transform for:

  • line detection in edge maps,
  • (optionally) circle detection.

• Simple object/shape detection pipeline based on edges + Hough transform.

All code for this homework is in hw2/TUZOV_ARTYOM_HW2.ipynb.

---

Homework 3 – Monocular & Stereo Vision (hw3/)

Focus: depth from stereo, disparity, and basic monocular geometry.

The hw3/data/ folder contains:

• HumanL.jpg, HumanR.jpg – stereo pair with a person.
• L5.jpg … L25.jpg, R5.jpg … R25.jpg – pairs of left/right images from a stereo rig.

Main ideas and tasks:

• Working with a calibrated stereo camera pair (left/right images).
• Computing disparity maps using block matching / semi-global matching.
• Converting disparity to depth and estimating distances to selected points in the scene.
• Visualizing epipolar geometry (corresponding points and epipolar lines).
• Using monocular cues and projection equations to reason about scale and distance.

All code for this homework is in hw3/TUZOV_ARTYOM_HW3.ipynb.

---

License

This repository is distributed under the terms of the license specified in LICENSE. Unless explicitly stated otherwise, the license covers only the code and documentation, not any third-party datasets or course materials.