AN IMPLEMENTATION PAPER ON A FRAMEWORK AND TECHNIQUES FOR IMAGE-BASED SEARCH APPLICATION WITH AN E-COMMERCE DOMAIN

Abstract

A large image database is growing rapidly as billions of images transferred every day at various chronicles like Instagram, snapchat, Facebook etc. and expected to continue in the future. With a particularly immense measure of data, the requirement for a successful pursuit in the image ages. Furthermore, if astoundingapparatuses have effectively been made for text search, image search stays an uncertain issue. Traditionally, most end-users use retrieval systems to write questions or queries and get text results. But endusers continuously expect search engines to be "intelligent" and they want a retrieval system to explore cyberspace using an image, as with the built-in camera on a mobile phone to send relevant resemblance, videos, and other formats. So, it becomes a major challenge to retrieve and processing images from this large database. Our paper introduced a novel methodology in contentbased image retrieval (CBIR) by consolidating the low level component for example shape, texture and color features. CBIR helps in retrieving images from an huge database in an efficient manner it is fastest growing research area in this field. The study found that convolutional neural networks can use to break down divisions and retrieval issues. Image details are much bigger than text data, and we cannot identify ocular particular by old techniques designed to identify text details. Therefore, CBIR has acquired significant benefits from research society. In this paper we aim to frame the feature vectors of the entire image by consolidating the shape, texture and color features and to get a decent performance in terms of the precision and recall. The image retrieval system is used to find images as per user request from the database. In this paper we also explored the flutter framework with dart language. We aim to mitigate the fundamental issue that app developers have been facing for so long- maintaining multiple apps for multiple platforms.