Recent advances in deep learning based computer vision

Deep learning is a hot topic in current AI research and significant progress has been made in a variety of fields, including computer vision and natural language processing. Traditional architecture of deep learning normally based on convolutional neural networks. In recent years, impressive results from Transformer models on natural language tasks have intrigued the vision community to study their application to computer vision problems. Thus, neural networks with Transformer architecture have been introduced to computer vision tasks and even challenge mainstream of traditional convolutional neural network. In this paper, it will review recent improvements in computer vision based on deep learning method. It'll go over some of the most important research in deep learning models. After that, it focus on advances in four key computer vision tasks: picture classification, object identification, semantic segmentation, and human posture estimation. It'll go through the most recent developments in these fields. Finally, it will provide a conclution of recent computer vision advances and discuss future possibility.