An Efficient CNN a deep learning approach applied on the image matching context

Abstract

Image matching is a quite challenging task to identify matching images in the data. There are multiple methods in computer vision techniques such as histogram-based algorithms, colour or edge based algorithms, textual based features, SIFT and Surf algorithms which will help to identify similar images. Here in our paper we are addressing an industrial problem to provide the better solution where US multinational courier delivery service facing challenges in delivering the products where labels/tags and bar codes of the products are missed while delivering to the customers and customers comes with the product image and with some information about the product. The job is to map the user or customer product information with the existing missed products. The advances in computer science and availability of GPU Machines, the problem will be addressed, and solutions can be automated using deep learning approaches. The paper describes the solution of matching the solution accurately and comparing the solution with the existing classical computer vision algorithms.