Md. Alimoor Reza
email: mreza AT gmu.edu

CS 682: Computer Vision
Course website


Image based retrieval

Attached code: image_based_retrieval.m computeAssignment.m

Vocab size	Accuracy rate	Saved mat file
500	60.61%	Saved mat file
800	59.09%	Saved mat file
1000	57.58%	Saved mat file
1200	48.48%	Saved mat file
1500	50.00%	Saved mat file
2000	40.91%	Saved mat file

In the developed bag of features model for image based retrieval, I have used various visual vocabulary sizes as tabulated above. I have used all the SIFT descriptors extracted from the image in all the experiments. I used the vlfeat library for SIFT feature extraction and vocabulary creation using k-means clustering.


Additional comments