Identifying States of Cooking Objects Using VGG Network

Abstract

Object recognition has been a long time problem since computer vision started to develop.In robotic learning, recognizing the state of an object for the robot is very important.In this paper, we are going to discuss the architecture of the VGG deep learning network to help with the state recognition problem since state recognition in nature is an image classification problem.The main work of this paper is to midify the structure of VGG to investigate if the accuracy of state classification could be improved together with some other modifications on the convolutional layers as well as the fully connected layers.