|
a |
|
b/CRCNet/README.md |
|
|
1 |
PyTorch code used to train CRCNet. |
|
|
2 |
|
|
|
3 |
Code brought from https://github.com/pytorch/vision/tree/master/references/classification. |
|
|
4 |
|
|
|
5 |
To use focal loss for training, you need to install `pytorch_toolbelt`: |
|
|
6 |
``` |
|
|
7 |
pip install pytorch_toolbelt |
|
|
8 |
``` |
|
|
9 |
|
|
|
10 |
See `train.sh` for how to train on your own data sets. An example of `trn.csv` and `val.csv` |
|
|
11 |
looks like: |
|
|
12 |
|
|
|
13 |
``` |
|
|
14 |
"image_name","tags" |
|
|
15 |
"/path/007.JPG.jpg","benign" |
|
|
16 |
"/path/008.JPG.jpg","malignant" |
|
|
17 |
"/path/009.JPG.jpg","benign" |
|
|
18 |
``` |
|
|
19 |
|