Rina Bao (ctc682) [email protected] The codes contain the implementations of our below paper, please cite our paper when you are using the codes.

Bao R, Al-Shakarji NM, Bunyak F, Palaniappan K. DMNet: Dual-stream marker guided deep network for dense cell segmentation and lineage tracking. In IEEE International Conference on Computer Vision (ICCV) Workshop on Computer Vision for Automated Medical Diagnosis, pp. 3354-3363, (2021).

This codes is for ISBI2021 6th Cell Segmentation and Tracking Challenge secondary track

Using Anaconda 3 (or miniconda3) on linux, run the following:

The Anaconda environment was tested on Linux with CUDA 10.2.

conda env create -f environment.yml
conda activate cell

1> Download the pretrained HRNet on imagenet from their website.

cd training_codes

mkdir models_imagenet

https://github.com/HRNet/HRNet-Image-Classification/

download the HRNet-W32-C model and put the model in models_imagenet

2> For six settings in "GT", "ST", "GT+ST", "allGT", "allST", "allGT+allST":

cd generate_bash

bash allGT.sh bash allST.sh bash allGT+ST.sh

For each dataset configuration training,

bash $dataset.sh

cd inference_codes

Running all bash files for testing

Thanks!