Extended Portfolio

Computer Vision

Core computer vision work spanning augmented reality, tracking, detection, and classical CV pipelines.

Markerless AR pipeline with feature extraction and matching, homography estimation, pose estimation, and camera calibration.
Marker-based AR with denoising, corner detection, KMeans clustering, and homography-based inverse warping.
Standard and hierarchical optical flow implementations using image pyramids and Lucas-Kanade methods.
Detection experiments using Hough transforms, Canny edges, and simulated/real-world traffic imagery.
Tracking pipelines using Kalman filters and particle filter approaches.
Classification
Face detection and recognition studies using PCA, boosting, Haar features, and Viola-Jones style techniques.