-
Notifications
You must be signed in to change notification settings - Fork 2
/
Readme.txt
30 lines (12 loc) · 1.08 KB
/
Readme.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
***********************************************************************************************************
Training code for "Single Infrared Image Stripe Noise Removal Using Deep Convolutional Networks"
***********************************************************************************************************
Usage:
1. Place the "Train_SNRCNN" folder into "($Caffe_Dir)/examples/" and modify the folder name as "SNRCNN"
2. Open MATLAB and direct to ($Caffe_Dir)/example/SNRCNN, run
"generate_train.m" and "generate_test.m" to generate training and test data.
3. To train our SNRCNN, run
./build/tools/caffe train --solver examples/SNRCNN/SNRCNN_solver.prototxt
4. After training, you can extract parameters from the caffe model and save them in the format that can be used in our test package (SNRCNN_Matlab). To do this, you need to install mat-caffe first, then open MATLAB and direct to ($Caffe_Dir) and run "saveFilters.m". The "($Caffe_Dir)/examples/SNRCNN/x1.mat" will be there for you.
## Contact with me
If you have any questions, please contact with me. (1007642157@qq.com)