Image segmentation is a topic of critical importance in the field of quantitative analysis of 2D and 3D images. Segmentation involves partitioning an image into multiple classes of objects or "segments" that are present in the images and subsequently determining the object boundaries with the best possible accuracy. Image segmentation tasks are often essential components of applications in fields such as computer vision, robotics, remote sensing, and computational biology.

In the last 5 years, deep learning models have received a massive focus and success in and as algorithms for image segmentation. Some marked benefits of deep learning based approaches include high accuracy, automatic and unsupervised operation, adaptability to new data, the capability to segment complex shapes, and robust operability, even for images of non-optimal qualities. This has generated huge interest in research for building advanced deep learning architectures for segmentation, as well as expanding the application areas of deep learning based segmentation in scientific, industrial, and consumer electronics domains.

This collection aims to bring together new developments in deep learning based image segmentation and its applications in fields like autonomous driving, robotic vision and navigation, augmented reality, biomedical image analysis, biometrics, video surveillance, and remote sensing. Of particular interest are new deep learning architectures for 2D and 3D image segmentation. Authors may also submit works on the creation of new training datasets with annotated ground truths, which are of immense importance in designing and training deep learning based segmentation models.