Deep Learning Techniques for Image Retrieval Applications

The increased use of digital computers, multimedia, and storage systems over recent years has result in large image and multimedia content repositories. This huge amount of multimedia data is being used in many fields. That rapid growing has created an ongoing demand of retrieval images systems operating on a large scale. Content Based Image Retrieval (CBIR) is the procedure of automatically retrieving images by the extraction of their visual features. Deep learning has been successfully applied to many problems e.g., computer vision and pattern recognition, computer games, robots and self-driving cars, voice recognition and generation, music composition and natural language processing. Due the success of deep leaning, and the importance of feature extraction in CBIR systems, this FDP aims to impart the concepts of feature extraction techniques in CBIR and the Deep learning model especially CNN for learning feature representation in CBIR. The CNN learns how to extract relevant features from a given images database and then applies this information in image retrieval process. The concepts are explained with hands on session and the participants are given with worksheets to apply the concepts they learnt.