Comments (23)
First step, I run GenerateTrainingPatches.m
,Second step ,I run Demo_Train_model_64_25_Res_Bnorm_Adam.m
to train.But I find data_size is NAN.what's wrong with it?
then I should run Demo_Test_model_64_25_Res_Bnorm_Adam.m
?
from dncnn.
By the way,if I just want to validate your code,could I run Demo_test_DnCNN.m
directly? But PSNR and SSIM both are NAN.so,could you figure the problem?
from dncnn.
maybe I solved above problem,but I met the follow problem :
>> Demo_test_DnCNN
引用了不存在的字段 'dilate'。
出错 vl_simplenn (line 303)
'dilate', l.dilate, ...
出错 Demo_test_DnCNN (line 64)
res = vl_simplenn(net,input,[],[],'conserveMemory',true,'mode','test');
maybe I solved above problem,but
from dncnn.
See #11
from dncnn.
hello , I run Demo_test_DnCNN.m , But PSNR and SSIM both are NAN.so,could you figure the problem?
and run DnCNN_train ,the result have error and is Input factor is insufficient DnCNN_train(line 75) net = vl_simplenn_tidy(net); can you help me
from dncnn.
I dont have this problem. It seems like you might have some problem with your training. How bout your training error? it should reduce around 1.0.
from dncnn.
what is difference for 1.0 and 1.1? l run the 1.0 also have error.
from dncnn.
l first run Demo_test_DnCNN3.m, the result is -----------------------------------------------
----BSD68------Gaussian Denoising-----
Average PSNR is NaNdB
Average SSIM is NaN
----Set5-----Super-Resolution-----
Average PSNR is NaNdB
Average SSIM is NaN
----Set14-----Super-Resolution-----
Average PSNR is NaNdB
Average SSIM is NaN
----BSD100-----Super-Resolution-----
Average PSNR is NaNdB
Average SSIM is NaN
----Urben100-----Super-Resolution-----
Average PSNR is NaNdB
Average SSIM is NaN
----classic5------Deblocking-----
Average PSNR is NaNdB
Average SSIM is NaN
----LIVE1------Deblocking-----
Average PSNR is NaNdB
Average SSIM is NaN
from dncnn.
For the NAN problem in version 1.1, you should (1) add the path to 'train400' data, then (2) run 'data\GenerateTrainingPatches.m', so that it will generate the training data for you. The training data is storaged in 'data\TrainingPatches\imdb_40_128.mat'. And please check your training data size 'imdb_40_128.mat' whether you really have the data or not.
from dncnn.
thank you very much
from dncnn.
wait , what is (1), 1.1 add or 1.0 add?
from dncnn.
You should add it for TrainingCodes_v1.1 since it doesn't have 'Train400' images in their 'data' folder. The 'Train400' images path is located in 'TrainingCodes_v1.0' folder.
from dncnn.
OK, thank you
from dncnn.
Hi, I don't understand how the mat files in model are created. Can you please tell in what sequence the code should run?
from dncnn.
hello. first, you should run Demo_test_DnCNN3. if you use GPU, ni should remove "%net = vl_simplenn_tidy(net);" "%", and if have error, in Demo_test_DnCNN3 add your vl_setupnn path. then run GenerateData_model_64_25_Res_Bnorm_Adam.
from dncnn.
I am actually new to deep learning. I didn't get how the models are created as .mat files. We need to run some code and then save it as mat file. Where is this code to create model? Thank you.
from dncnn.
from dncnn.
I am new to deep learning. how did u you resolve the data_size and data_meme NaN issue? @goonder .
from dncnn.
也许我解决了上面的问题,但我遇到了以下问题:
>> Demo_test_DnCNN 引用了不存在的字段 'dilate'。 出错 vl_simplenn (line 303) 'dilate', l.dilate, ... 出错 Demo_test_DnCNN (line 64) res = vl_simplenn(net,input,[],[],'conserveMemory',true,'mode','test');
也许我解决了上面的问题,但是
您好,我现在也遇到了同样的问题,请问您是怎么解决的。
from dncnn.
I am new to deep learning. how did you resolve the data_size and data_meme NaN issue? @goonder @allenkate12
from dncnn.
from dncnn.
maybe I solved above problem,but I met the follow problem :
>> Demo_test_DnCNN 引用了不存在的字段 'dilate'。 出错 vl_simplenn (line 303) 'dilate', l.dilate, ... 出错 Demo_test_DnCNN (line 64) res = vl_simplenn(net,input,[],[],'conserveMemory',true,'mode','test');
maybe I solved above problem,but
How can you solve this problem,could you tell me? I have met the same problem,my datasiaze is also NAN.
from dncnn.
from dncnn.
Related Issues (20)
- 有训练好的Pytorch的彩色去噪模型吗
- Color Images input as test in keras version
- About training data creating HOT 1
- color dcnn model's act_mode is different in train and test phase (pytorch version)
- Size of input images
- Version of DnCNN that is compatible with tensorflow 2.0
- How to resolve this bug? could you please tell me?maybe I solved above problem,but I met the follow problem :
- bug in usage of OpenCV resize function
- Batch Normalization gives worse perfromence
- ERROR: Could not find a version that satisfies the requirement torch==0.4.1 ERROR: No matching distribution found for torch==0.4.1 HOT 1
- Data Preparation for DnCNN3
- vl_argparse(opts, varargin);No function or variable is defined—— 'vl_argparse'。And I can't find matlab code about this function
- Are there any files missing?Such as vl_simplenn_move.m and vl_argparse.m
- main_test.py
- Sigma=45 model in .h5
- If I want to input my own image for deep learning,but i don't need add to Gassian noise level,what should I do? HOT 1
- Lower depth yields same performance
- Repetitive Dataset and DataLoader Creation in PyTorch version
- About Testing
- About image shape debug HOT 2
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from dncnn.