This project aims to explore and evaluate the performance of deep CNNs with Resnet and VGG Net architectures for image classification tasks. We conduct experiments on standard benchmark datasets and compare the results with state-of-the-art methods to analyze the strengths and weaknesses of the models. Furthermore, we provide insights into the internal representations of the deep CNNs and their effectiveness in handling challenging tasks such as fine-grained classification. Overall, our findings demonstrate the potential of deep CNNs for image classification tasks and highlight the importance of continued research in this area.