The neural network architecture was inspired by U-Net: Convolutional Networks for Biomedical Image Segmentation and based on the previous work Convolutional neural networks for automated annotation of cellular cryo-electron tomograms

---

Model

Loss function:

---

How to use

1. Prepare pre-processed data and label as described.
2. Use main.py for trainning and testing.
3. Use predict.py on new dataset.

Dependencies

• Pytorch