Enhancing CNN Model by Incorporating Upscaling and Downscaling Techniques for Image Classification

Image classification is a fundamental task in computer vision with numerous applications. Even though, the rapid development of Convolutional Neural Networks (CNNs) has revolutionized the field of image classification, there is a constant pursuit for further improvement. Existing CNN models typically rely solely on downscaling techniques for pre-processing, limiting their potential for extracting intricate image details and effectively handling varying image resolutions. In this paper, we propose an approach that incorporates both upscaling and down-scaling techniques to enhance the performance of CNN models for image classification. Our method leverages the benefits of both upscaling and downscaling operations to capture fine-grained details and extract high-level features, respectively. Through extensive experiments on benchmark datasets, namely CIFAR-10 and CIFAR-100 dataset, we demonstrate the superiority of our proposed approach over existing methods in terms of accuracy and robustness. For instance, our proposed model, named Super-Res VGG has received 85.57% testing accuracy whereas VGG16, GoogLeNet (Inception), and ResNet50 have received 84.40%, 73.93%, and 65.29% accuracy on CIFAR-10 dataset, respectively.