- Opencv
- Python 3.6
- Tensorflow
- Keras (Tf backend)
The data used for training is from https://www.kaggle.com/c/data-science-bowl-2018/data
- Simply run nuclei-segmentation.py
- Image should be in .png format
- Accuracy is about 94.8%
Unet is an auto-encoder used with medical image. I have used it for semantic segmentation and results are quite good.