Old projects from Computer Vision
Proj1: Creating hybrid images by using a low pass and high pass gaussian kernel.
Proj2: Detecting interest points using harris corner detecter and then gathering SIFT features to match interest points in images of the same scene/object from different view points.
Proj3: Scene classification using bag of words model, k-nearest neighbors and clustering. I also implemented spatial pyramid matching to boost accuracy.
Proj4: Facial recognition using a sliding window technique and a linear SVM.
Proj5: Implemented the pipeline in the sketch tokens paper. This used a sliding window technique, clustering, and a random forest classifier. I was able to beat canny edge detector and sobel edge detector.