A Survey of Research Progresses on Instance Segmentation Based on Deep Learning

Instance segmentation is a crucial research task in the field of computer vision, providing an indispensable theoretical foundation for the development of panoramic segmentation technology. It has a wide range of applications in many areas, such as cyber security, intelligent driving, tumor recognition, boundary segmentation, pest, disease recognition, face recognition and beauty enhancement, etc. With the continuous development of deep learning, instance segmentation technology has also been continuously improved, and the segmentation accuracy has been increasing. This survey provides a comprehensive review of the latest instance segmentation methods, including the improvement of existing convolutional neural networks, innovative segmentation networks, adaptive learning, and so on. At the same time, the data set was summarized and the performance of the above methods was summarized. The existing application scenarios and future research and development directions were also discussed.