Image Classification using Deep Learning Models in Computer Vision

This research focuses on comparison of deep learning architectures employed in computer vision specifically image classification, for classification we employ three cutting edge deep learning architectures: Xception, DenseNet, and the Vision Transformer (ViT).We evaluate and compare the performance of these models in accurately classifying images of Birds. This comprehensive study aims to provide insights into the effectiveness of different deep learning models, shedding light on their strengths and weaknesses in image classification. This research delves into the exploration of ImageNet21k, a subset featuring 10 carefully choosen classes of Birds, ImageNet stands out as a cornerstone, serving as a benchmark for evaluating the capabilities of deep neural networks and a huge dataset with millions of labelled images spread over thousands of categories, has become the touchstone for testing the efficacy of deep learning models. The task involves not only accurately classifying objects within images but also detecting and localizing multiple objects, pushing the boundaries of artificial intelligence in visual comprehension.