Human, Object and Scene Centric Image Retrieval Engine to Enhance Image Management

Abstract

Image data available on internet and in personal computers is colossal. There is a need of a search engine that can effectively meet the retrieval demands of user. Most of the systems available consider low level features for retrieval without taking input from user. To handle this problem, we propose a search engine that can retrieve images from database based on specific request from user. We present a system that has multiple computer vision algorithms based retrieval options available. Scene, human, face, age, emotion, face recognition, gender and object detection based systems are integrated to create a diverse image search engine. The retrieval performance of system is shown in pictorial form. Precision and recall metrics are used to evaluate system’s performance.