A Comprehensive Analysis on Deep Learning based Image Retrieval

Image retrieval is a computer vision task that entails looking for images that resemble a particular query image in a sizable database. By using visual similarity or other factors, image retrieval seeks to empower users to locate images that correspond to their interests or requirements. The image processing stream is recognized as an essential module since a lot of systems use logic to map inputs to outputs. Semantic gaps prevent some conventional text-and content-based picture retrieval techniques from reflecting the human perception of images. The primary aim of this paper is to provide a research survey on traditional Deep Learning based image retrieval techniques along with their advantages, disadvantages, and limitations. This study focuses on the issues in Content Based Image Retrieval (CBIR), such as the semantic gap that develops among image pixels with a low level (collected by machines) and image pixels with a high level (collected by humans). The extensive approach supports researchers to attain greatest solutions for the present issues in CBIR. The accuracy, precision, recall, and f-score are measured key parameters for determining the efficiency of image retrieval techniques.