The DeepLab Family

Image segmentation tasks have seen lots of developments in recent years, and have become one of the most researched topics in Computer Vision⁶. One of the standards for segmentation is represented by the Deep Labelling for Image Segmentation architecture, also known as DeepLab. The approach was developed by Chen et al.¹ ² ³ ⁴ and different versions employing different mechanisms were proposed over time. In this article, a brief overview of the different DeepLab algorithms and their basic functioning will be given. The first appearance of the DeepLab architecture is found in [1].