A Survey of Popular Image and Text analysis Techniques

Abstract

Image processing has made huge progress in recent times and has captured the attention of the international research community in recent times. In this survey paper the winning Convolutional Neural Network (CNN) architectures of the popular ImageNet Large Scale Visual Recognition Competition (ILSVRC) competition along with the innovations they introduced are discussed in detail. CNNs to the likes of the ZFNet, Inception V1, V2, V3 and V4 versions, ResNet versions, Inception ResNet versions, VGG versions are all covered. Also object detection is looked at as a preeminent task and popular models like RCNN, Fast RCNN, Faster RCNN and the YOLO versions along with their award winning architectures are discussed. Text classification is another major practice that is followed before making business decisions. In this survey paper section III explains the text classification techniques. preprocessing techniques such as removing stop words, stemming, lemmatization. Two variants of vectorization method such as vectorization-delta, vectorization-center. paper also explains the hyperparameter tuning for optimization of support vector machine(SVM), Logistic regression(LR), random forest(RF) and Boosted Regression Tree (BRT). Explored fine tuning parameter which enhance the performance of machine learning models.