Comments (8)
ThorstenFalk
commented on July 23, 2024
1
OK, that's the structure of the .modeldef.h5 file. I don't know how you managed to overwrite it, but you probably have to download the model again (or fetch the .caffemodel.h5 file from the zip-archive if you still have it in your downloads)
from unet-segmentation.
ThorstenFalk
commented on July 23, 2024
Running the backend on Windows is currently not supported. On CentOS, you can build caffe_unet the same way you build vanilla caffe. I have no CentOS machine for testing it, but the caffe yum installation guide should be a useful resource.
from unet-segmentation.
ech3000
commented on July 23, 2024
I managed to install it on Centos 7, via OpenBLAS.
However, when testing it via the FIJI plugin i.e. doing the simple example segmentation. I'm getting a "Model/Weight check failed" error. It didn't give me more infos, so ran the command in the terminal.
Here's what it says:
$ caffe_unet check_model_and_weights_h5 -model /home/me/U-net_Projekt/caffemodels/2d_cell_net_v0.caffemodel.h5 -weights /home/me/U-net_Projekt/caffemodels/2d_cell_net_v0.caffemodel.h5 -n_channels 1 -gpu 0
I0423 14:33:38.333029 19345 caffe_unet.cpp:172] Checking model /home/me/U-net_Projekt/caffemodels/2d_cell_net_v0.caffemodel.h5 and weights /home/me/U-net_Projekt/caffemodels/2d_cell_net_v0.caffemodel.h5
I0423 14:33:38.333822 19345 caffe_unet.cpp:179] Use GPU with device ID 0
I0423 14:33:38.426545 19345 caffe_unet.cpp:183] GPU device name: ��Q�
F0423 14:33:38.426599 19345 common.cpp:152] Check failed: error == cudaSuccess (30 vs. 0) unknown error
*** Check failure stack trace: ***
@ 0x7f1b8878de6d (unknown)
@ 0x7f1b8878fced (unknown)
@ 0x7f1b8878da5c (unknown)
@ 0x7f1b8879063e (unknown)
@ 0x7f1b8995657a caffe::Caffe::SetDevice()
@ 0x408ab6 check_model_and_weights_h5()
@ 0x407098 main
@ 0x7f1b7b57a3d5 __libc_start_main
@ 0x407891 (unknown)
Aborted (core dumped)
Specs:
OS: Centos 7
GPU: Nvidia Quadro K2200
Cuda 10.0
GPU Driver: 410.78
some cuDNN version is installed, but I didn't use it during installation (left it commented out in the ~/caffe/Makefile.config). This shouldn't cause any issues, right?
So the driver and the cuda versions should be compatible. But why am I getting this error? Also why can't it output my GPU device name?
Since this is clearly GPU related I've tried to run it with CPU only. This causes a Input/Output error:
java.io.IOException: Error during segmentation: exit status 134
Unknown caffe error.
Here's a file with a copy of the Log output.
Java_IOerror.txt
Any idea what this is all about?
All help is very much appreciated.
Best,
Reto
PS: I apologize if you'd rather have this post as a new issue.
from unet-segmentation.
ech3000
commented on July 23, 2024
and now with a txt file where it says more than 'asdf'
classic mistake xD
from unet-segmentation.
ThorstenFalk
commented on July 23, 2024
The error on CPU occurs when loading the weights. My first guess is that you accidentally uploaded the .modeldef.h5 file instead of the caffemodel.h5 file, when you were asked to upload the weights.
The other error looks strange. The GPU identifier string seems to be garbage, which indicates some memory issue. I will try to reproduce the problem.
from unet-segmentation.
ech3000
commented on July 23, 2024
That might be the case. How do I change that?
I'm running both back and frontend on the same machine, by the way.
Thanks for the help.
from unet-segmentation.
ThorstenFalk
commented on July 23, 2024
Do you have hdf5 utility functions installed? If so, please check the output of h5ls -r /home/wilret00-adm/U-net_Projekt/caffemodels/2d_cell_net_v0.caffemodel.h5
It should look like this:
/ Group
/data Group
/data/augm_data2-data3 Group
/data/concat_d0c_u0a-b Group
/data/concat_d1c_u1a-b Group
/data/concat_d2c_u2a-b Group
/data/concat_d3c_u3a-b Group
/data/conv_d0a-b Group
/data/conv_d0a-b/0 Dataset {64, 1, 3, 3}
/data/conv_d0a-b/1 Dataset {64}
/data/conv_d0b-c Group
/data/conv_d0b-c/0 Dataset {64, 64, 3, 3}
/data/conv_d0b-c/1 Dataset {64}
/data/conv_d1a-b Group
/data/conv_d1a-b/0 Dataset {128, 64, 3, 3}
/data/conv_d1a-b/1 Dataset {128}
/data/conv_d1b-c Group
/data/conv_d1b-c/0 Dataset {128, 128, 3, 3}
/data/conv_d1b-c/1 Dataset {128}
/data/conv_d2a-b Group
/data/conv_d2a-b/0 Dataset {256, 128, 3, 3}
/data/conv_d2a-b/1 Dataset {256}
/data/conv_d2b-c Group
/data/conv_d2b-c/0 Dataset {256, 256, 3, 3}
/data/conv_d2b-c/1 Dataset {256}
/data/conv_d3a-b Group
/data/conv_d3a-b/0 Dataset {512, 256, 3, 3}
/data/conv_d3a-b/1 Dataset {512}
/data/conv_d3b-c Group
/data/conv_d3b-c/0 Dataset {512, 512, 3, 3}
/data/conv_d3b-c/1 Dataset {512}
/data/conv_d4a-b Group
/data/conv_d4a-b/0 Dataset {1024, 512, 3, 3}
/data/conv_d4a-b/1 Dataset {1024}
/data/conv_d4b-c Group
/data/conv_d4b-c/0 Dataset {1024, 1024, 3, 3}
/data/conv_d4b-c/1 Dataset {1024}
/data/conv_u0b-c Group
/data/conv_u0b-c/0 Dataset {128, 192, 3, 3}
/data/conv_u0b-c/1 Dataset {128}
/data/conv_u0c-d Group
/data/conv_u0c-d/0 Dataset {128, 128, 3, 3}
/data/conv_u0c-d/1 Dataset {128}
/data/conv_u0d-score Group
/data/conv_u0d-score/0 Dataset {2, 128, 1, 1}
/data/conv_u0d-score/1 Dataset {2}
/data/conv_u1b-c Group
/data/conv_u1b-c/0 Dataset {128, 256, 3, 3}
/data/conv_u1b-c/1 Dataset {128}
/data/conv_u1c-d Group
/data/conv_u1c-d/0 Dataset {128, 128, 3, 3}
/data/conv_u1c-d/1 Dataset {128}
/data/conv_u2b-c Group
/data/conv_u2b-c/0 Dataset {256, 512, 3, 3}
/data/conv_u2b-c/1 Dataset {256}
/data/conv_u2c-d Group
/data/conv_u2c-d/0 Dataset {256, 256, 3, 3}
/data/conv_u2c-d/1 Dataset {256}
/data/conv_u3b-c Group
/data/conv_u3b-c/0 Dataset {512, 1024, 3, 3}
/data/conv_u3b-c/1 Dataset {512}
/data/conv_u3c-d Group
/data/conv_u3c-d/0 Dataset {512, 512, 3, 3}
/data/conv_u3c-d/1 Dataset {512}
/data/create_deformation Group
/data/d0c_relu_d0c_0_split Group
/data/d1c_relu_d1c_0_split Group
/data/d2c_relu_d2c_0_split Group
/data/d3c_dropout_d3c_0_split Group
/data/def_create_deformation_0_split Group
/data/def_data-data2 Group
/data/def_label-crop Group
/data/def_weight-crop Group
/data/dropout_d3c Group
/data/dropout_d4c Group
/data/loaddata Group
/data/loss Group
/data/pool_d0c-1a Group
/data/pool_d1c-2a Group
/data/pool_d2c-3a Group
/data/pool_d3c-4a Group
/data/relu_d0b Group
/data/relu_d0c Group
/data/relu_d1b Group
/data/relu_d1c Group
/data/relu_d2b Group
/data/relu_d2c Group
/data/relu_d3b Group
/data/relu_d3c Group
/data/relu_d4b Group
/data/relu_d4c Group
/data/relu_u0a Group
/data/relu_u0c Group
/data/relu_u0d Group
/data/relu_u1a Group
/data/relu_u1c Group
/data/relu_u1d Group
/data/relu_u2a Group
/data/relu_u2c Group
/data/relu_u2d Group
/data/relu_u3a Group
/data/relu_u3c Group
/data/relu_u3d Group
/data/reshape_data Group
/data/reshape_labels Group
/data/reshape_weights Group
/data/reshape_weights2 Group
/data/trafo_data3-d0a Group
/data/upconv_d4c_u3a Group
/data/upconv_d4c_u3a/0 Dataset {1024, 512, 2, 2}
/data/upconv_d4c_u3a/1 Dataset {512}
/data/upconv_u1d_u0a Group
/data/upconv_u1d_u0a/0 Dataset {128, 128, 2, 2}
/data/upconv_u1d_u0a/1 Dataset {128}
/data/upconv_u2d_u1a Group
/data/upconv_u2d_u1a/0 Dataset {256, 128, 2, 2}
/data/upconv_u2d_u1a/1 Dataset {128}
/data/upconv_u3d_u2a Group
/data/upconv_u3d_u2a/0 Dataset {512, 256, 2, 2}
/data/upconv_u3d_u2a/1 Dataset {256}
from unet-segmentation.
ech3000
commented on July 23, 2024
$ h5ls -r /home/wilret00-adm/u-net/caffemodels/2d_cell_net_v0.caffemodel.h5
/ Group
/.unet-ident Dataset {SCALAR}
/model_prototxt Dataset {SCALAR}
/solver_prototxt Dataset {SCALAR}
/unet_param Group
/unet_param/description Dataset {SCALAR}
/unet_param/downsampleFactor Dataset {1}
/unet_param/element_size_um Dataset {2}
/unet_param/input_blob_name Dataset {SCALAR}
/unet_param/input_dataset_name Dataset {SCALAR}
/unet_param/mapInputNumPxGPUMemMB Dataset {2, 79}
/unet_param/name Dataset {SCALAR}
/unet_param/normalization_type Dataset {SCALAR}
/unet_param/padInput Dataset {1}
/unet_param/padOutput Dataset {1}
/unet_param/padding Dataset {SCALAR}
/unet_param/pixelwise_loss_weights Group
/unet_param/pixelwise_loss_weights/borderWeightFactor Dataset {SCALAR}
/unet_param/pixelwise_loss_weights/borderWeightSigma_um Dataset {SCALAR}
/unet_param/pixelwise_loss_weights/foregroundBackgroundRatio Dataset {SCALAR}
/unet_param/pixelwise_loss_weights/sigma1_um Dataset {SCALAR}
It looks like this.
from unet-segmentation.
Related Issues (20)
- Exception in thread "Thread-5" java.lang.IllegalArgumentException HOT 1
- Where to obtain test data?
- Tiling makes tens of thousands of images HOT 3
- Use the weights available at https://lmb.informatik.uni-freiburg.de/resources/opensource/unet/ for Tensorflow 2 (Keras) HOT 6
- Invalid Private Key, RSA HOT 4
- CUDA 11 support HOT 3
- How to change binary path if I am using EC2? HOT 2
- output very weird HOT 5
- CUDA 10.1 support HOT 3
- U-net installation HOT 6
- Finetuning does not 'train' ,aborts HOT 6
- Segmentation does not have good separation between multiple instances of cells
- Writing a macro script to run the Finetune model
- caffe_unet not found HOT 1
- Model/weight check failed
- client loop: send disconnect: broken pipe; SFTP Failure 4 HOT 1
- caffe or caffe_unet issues HOT 7
- MacOs installation issue HOT 1
- Exporting the environment
- Installation Roadblock HOT 1
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 unet-segmentation.