Content-Based Image Retrieval using SIFT and CNN

The graphical data is increasing rapidly and most of the data is in the form of images. The reason for this rapid increase in images is the digitalization of data in sectors like medical, education, commerce, government, etc. For searching through a pile of images most of the sectors still use primitive methods like searching manually or assigning keywords to the images. In this paper, we are trying to solve this problem by using CBIR (Content-Based Image Retrieval) technique. In CBIR we extract features from images and using those features we can match similar images which will be helpful in finding a particular image or set of images from the dataset. We are also using CNN for object detection which will help us in finding labels of objects and using those labels we can classify and store images for accelerating search. The main algorithm used for keypoints extraction is SIFT (Scale Invariant Feature Transform). We've also tested our project with Face Images dataset.