This repository contains all the codes related to the assignments and homeworks I did for the course Computer Vision (CSE344 / ECE344).
Welzl Algorithm- Assignment-1.py detects all the individual objects in grayscale image, finds a minimum bounding circle for each one of them and evaluates the accuracy of the bounding circle using Jaccard similarity. The minimum bounding circle is computed using Welzl algorithm.
CV_homeworks_notebook.ipynb contains the codes from most of the homeworks and a couple of assignments.
Contrast and Spatial cues-Assignment 2.py implements algorithm for finding contrast cue and spatial cue from an RGB image. Contrast cue and Spatial cue helps in Image co-saliency and co-segmentation. More can be read at https://www.researchgate.net/publication/236581375_Cluster-Based_Co-Saliency_Detection. It further obtain quality scores for those cues using a metric called "separation measure". Further can be read about separation measure at https://www.researchgate.net/publication/321796205_Quality-Guided_Fusion-Based_Co-Saliency_Estimation_for_Image_Co-Segmentation_and_Colocalization.
Region Segmentation-Assignment 2.py implements region segmentation on an RGB image using Fuzzy c-means algorithm.
Assignment-3.ipynb uses MNIST dataset to develop individual deep learning models that performs a) Foreground extraction b) Classification along with circlization c) Semantic segmentation Evaluation of all the predicted images on the test dataset is done using Jaccard similarity.