Deep-Sceen-Classification

Comparison of Sceen Classification accuracy between Histogram, SIFT and Deep Learning based features

Scene classification with using some certain different images of scenes are very important and crucial issue in computer vision literature. Especially, to automatize it with computer has significant amount of benefit in terms of robotic and automation. Although computers are still far from the human beings ability in order to visual understanding, the researchers have done too many significant contribution on this area. As general classification problem, image scene classification problem has the same two fundamental step in it. These are feature extraction and classification respectively. In this research, we have focused on three different image features which are histogram of color, bag of sift features and finally convolution neural networks. For classification method, we have just tried multi class support vector machine. Our research shows that pretrained convolutional neural network named vgg16 has reasonable accuracy.