Deep Inside Feature Learning for Image Classification Using Transfer Learning Approach

Abstract

A scene consists of manifold objects and some relations among them. Exploring the relation among objects to acquire a meaningful understanding find application in the field of remote sensing, robotics, self driving cars etc. Our study intend to explore one such scene activity that can assist in resciung during natural disasters for example, flood. We have used the concept of deep learning networks which is a sub class of machine learning to develop a model that can detect the vehicles which got drowned in flood. We have made our own small dataset consisting of four classes with 100 images relating vehicles in each class. In our work convolutional neural network based pretrained model resnet101 learn the features and Support Vector Machines(SVM) functions as the classifier. This approach has shown an overall success rate of 91% in classification.