Cell Segmentation with U-Net and Others

Deepcell[1] is a framework I worked on for a bit during my undergrad. It can do cell segmentation & tracking for different imaging types.

[1]: https://github.com/vanvalenlab/deepcell-tf/blob/master/READM...

For this particular problem, Mask R-CNN would have been the way to go -- it spits out instances as opposed to just deciding, for each pixel, to which class it belongs. Or an SSD (if we don't care about the mask at all).

U-net is great and there are plenty of different versions of it for a variety of problems that have yet to be discovered!