2019/07/26 by Bulla Rajesh, Muhammad Faisal Javed, Rajesh, Bulla +4
Computer Science · Medicine · Neuroscience · #Advanced Neural Network Applications #Brain Tumor Detection and Classification #COVID-19 diagnosis using AI #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Machine Learning (cs.LG)
paper · pdf · doi:10.48550/arxiv.1907.11503
openalex publication_date 2019/07/26 · openalex created_date 2022/07/28 · openalex updated_date 2026/07/28
The popularity of Convolutional Neural Network (CNN) in the field of Image\nProcessing and Computer Vision has motivated researchers and industrialist\nexperts across the globe to solve different challenges with high accuracy. The\nsimplest way to train a CNN classifier is to directly feed the original RGB\npixels images into the network. However, if we intend to classify images\ndirectly with its compressed data, the same approach may not work better, like\nin case of JPEG compressed images. This research paper investigates the issues\nof modifying the input representation of the JPEG compressed data, and then\nfeeding into the CNN. The architecture is termed as DCT-CompCNN. This novel\napproach has shown that CNNs can also be trained with JPEG compressed DCT\ncoefficients, and subsequently can produce a better performance in comparison\nwith the conventional CNN approach. The efficiency of the modified input\nrepresentation is tested with the existing ResNet-50 architecture and the\nproposed DCT-CompCNN architecture on a public image classification datasets\nlike Dog Vs Cat and CIFAR-10 datasets, reporting a better performance\n