Abstract

In this research, we propose a novel visual task named instance image classification to achieve the instance based image classification that has not been achieved easily in otherwise using current imaging or computer vision methods. The research involves the use of projection a laser on an object. The projected laser pattern is extracted from the image using Faster RCNN object detection detector. The extracted image is further subjected to a Inception V3 network for classification. Our results show that the proposed technique is able to achieve high accuracy of instance based image classification. The proposed method has a wide range of applications for self-driving, robotics, and automation.

© 2020 The Author(s)