Computer Vision

● Deployed visual object tracking with Lucas-Kanade forward additive image alignment algorithm and Matthew-Bakers Inverse Compositional alignment algorithm.
● Performed 3-D Reconstruction using bundle adjustment and estimated Fundamental matrix with 8/7-point algorithm and RANSAC.
● Worked on stitching panoramas utilizing homography. Performed OCR from handwritten letters with a CNN model trained on MNIST dataset.
● Implemented the Hough Transform to detect lines in an image.
● Extracted SIFT features to build a Bag-of-Words representation of an image for classification.

Learning Outcomes:

1. Hough Transform
2. Bag of Visual words
3. Neural Networks
4. Homorgraphy and RANSAC
5. Camera Calibration, Triangulation, Epipolar Geometry, Rectification and 3D Reconstruction
6. Lucas Kanade and Mathew Bakers Image alignment

Programming Language: Pytorch, Python