Two Phase Content Based Image Retrieval with relevance ranking on classified images

With the growing trend in mobile and computing devices, large number of digital images are generated in day to day life. These generated images are stored in databases for later use. As enormous quantity of digital images are stored in database there is a need for retrieving the relevant image. The content - based image recovery (CBIR) that manages the recovery of images from the large database is one of the applications of computer vision. Content - based image recovery is carried out in two phases in this research work. In the first phase Convolutional Neural Network(CNN) is used for detecting the images. CNN are capable of detecting multiple objects each belonging to different class by processing the image in a single pass. Convolution Neural Network is used to solve the multiple class detection problem. In the second phase once the objects are detected, relevant images are retrieved by using relevance ranking where priority is assigned within same class of images.