Advanced Computer Vision

This repository contains my solutions for Advanced Computer Vision (EE243) course homeworks/projects offered by Prof. Amit K. Roy-Chowdhury, University of California, Riverside, Spring 2019.

Directories

Each homework directory contains:

• problem directory: includes the definition of the problem, independent reading files, and template code files.
• solution directory: includes a report pdf file as well as the solution implementation (if asked by the problem definition).

Overview

• Homework 1: problem definition, report.

  • Intro to image processing with Matlab, DCT, DFT, Noise, De-noise.

• Homework 2: problem definition, report

  • Obtaining a multi-resolution decomposition up to two levels of scale using Haar wavelet.
  • Reconstructing back the original image using all the multi-resolution decompositions.
  • Edge detection using Laplacian of the Gaussian and Canny edge detector.
  • Line detection using Hough transform.
  • Implementation of Shi-Tomasi corner detector.
  • Feature matching (HoG and SIFT features).

• Homework 3: problem definition, report

  • Implementation of basic version of normalized cuts for segmenting.
  • Implementation of the Expectation Maximization algorithm for mixture of Gaussian model based on color features for segmenting.

• Homework 4: problem definition, report

  • Feature extraction
  • Logistic Regression

• Homework 5: problem definition, report

  • Training a Convolutional Neural Network (CNN) from scratch using PyTorch.

• Homework 6: problem definition, report

  • Homography Estimation
  • Implementation of the factorization based Structure from Motion (SFM) method.
  • Fundamental Matrix Estimation.

• Homework 7: problem definition, report

  • Problem set on epipolar geometry, camera model/calibration, stereo reconstruction.