Local-Feature-Matching

Local Feature Matching: Computer Vision University Project

This project aims at learning to generate image features around a local point in the image.

The project consists of three parts in student.py:

1. Generating Interest Points with Harris get_interest_points
2. Generating SIFT-like features around each interest point get_features
3. Matching Features between two images match_features

Results:

Notre Dame

Matches after Whole pipeline on Notre Dame. Matches: 1113 Accuracy on 50most confident: 100% Accuracy on 100 most confident: 99% Accuracy on all matches:75%

Mt Rushmore

Matches after Whole pipeline on Mount Rushmore . Matches: 55 Accuracy on50 most confident: 94% Accuracy on all matches: 92%

E Gaudi

Matches after Whole pipeline on Epicopal Gaudi . Matches: 13 Accuracy onall matches: 23%