UNet is the most commonly used and simplest segmentation model. It is simple, efficient, easy to understand, easy to build, and can be trained from small datasets. The original intention of UNet is to solve the problem of medical image segmentation. In terms of solving the task of cell-level segmentation, it won multiple firsts in the ISBI cell tracking competition in 2015. After that, UNet has been widely used in various directions of semantic segmentation (such as satellite image segmentation, industrial defect detection, etc.) due to its outstanding segmentation effect.
The following table lists the 2DUnet model supported by the Vitis AI Library.