I am getting the following error when I am trying to extract features using vlc_reference_caffenet.caffemodel
./build/tools/extract_features.bin models/bvlc_reference_caffenet/bvlc_reference_caffenet.caffemodel examples/_temp/imagenet_val.prototxt fc7 examples/_temp/features 10 1
E0102 21:58:00.103168 23134 extract_features.cpp:62] Using CPU
F0102 21:58:00.664142 23134 upgrade_proto.cpp:75] Check failed: ReadProtoFromBinaryFile(param_file, param) Failed to parse NetParameter file: models/bvlc_reference_caffenet/bvlc_reference_caffenet.caffemodel
* Check failure stack trace: *
@ 0x7f0e01d69daa (unknown)
@ 0x7f0e01d69ce4 (unknown)
@ 0x7f0e01d696e6 (unknown)
@ 0x7f0e01d6c687 (unknown)
@ 0x7f0e0210624e caffe::ReadNetParamsFromBinaryFileOrDie()
@ 0x7f0e0214e137 caffe::Net<>::CopyTrainedLayersFromBinaryProto()
@ 0x7f0e0214e1a6 caffe::Net<>::CopyTrainedLayersFrom()
@ 0x408d52 feature_extraction_pipeline<>()
@ 0x7f0e00f95ec5 (unknown)
@ 0x4048be (unknown)
@ (nil) (unknown)
Aborted (core dumped)
Hi ThakurRaj how do you solved it? I'm getting same
Check failed: ReadProtoFromBinaryFile(param_file, param) Failed to parse NetParameter file: ../models/bvlc_reference_caffenet/bvlc_reference_caffenet.caffemodel
when running second cell of 00-classification.ipynb sample!
I'll appreciate any help!
Cheers
Sorry I dont understand... could you put a link with some suggestions... thanks
@renexdev I exactly don't remember but it was some caffe version issue and they have changed some parameters name and I didn't change them .... it was kinda easy to solve and I did fix it very quickly. Try to see if parameters name you are using are correct otherwose I think I can send you the whole sample code using Kaggle competition data.
OK thanks, I'll check it. By the way I opened the downloaded bvlc_reference_caffenet.caffemodel file and is unreadable... it is supposed you can see the parameter structure... so I delete it and now I'm downloading the file again.. may be there was a connection problem...
I exactly don't remember. Give me a day and I will send you the whole code to run it successfully :)
Ok thanks Thakur! seems to work right now! the bvlc_reference_caffenet.caffemodel file sizes 230MB now, before was 30MB, so there was some problem with the internet connection. Is there any gitter.im group going around for the beginers in caffe? I'm quite lost with the whole zoo models and documentation! Thanks again for your help!
Is this issue solved? I'm facing the same problem.
Yeap @kshitijgoel007 in my case was a matter of the incomplete file downloading. I was facing internet connection issues at that time. Hope it helps. Cheers
Hi @renexdev I'm facing the same problem. I can get the "features" folder ,but there are still some errors. How can I solve them?
./build/tools/extract_features.bin models/bvlc_reference_caffenet/bvlc_reference_caffenet.caffemodel examples/_temp/imagenet_val.prototxt fc7 examples/_temp/features 10 leveldb
E0405 15:18:03.274474 18745 extract_features.cpp:62] Using CPU
E0405 15:18:04.010071 18745 extract_features.cpp:133] Extacting Features
^[[AE0405 15:18:11.823441 18745 extract_features.cpp:176] Extracted features of 500 query images for feature blob fc7
E0405 15:18:11.823612 18745 extract_features.cpp:181] Successfully extracted the features!
Hello @renexdev @kshitijgoel007 @nikoong @ThakurRajAnand I am facing this problem now. I my case, i just have extract_features.exe and extract_features.pdb. I do not have extract_features.bin. I am running in windows.
And every single time I try to run ./build/tools/extract_features examples/_temp/bvlc_reference_caffenet.caffemodel examples/_temp/imagenet_val.prototxt pool3 examples/_temp/features 10 leveldb I got ** Check failure stack trace: ** error.
Please help me. Thank you.
Sorry guys I'm not and expert in the topic. At that time it seems was a problem with my internet connection and I solved the issue removing the project with the generated files and downloading/ compiling the project again. Cheers.
@ThakurRajAnand Hello, I met the same problem when I trying to finetune the ResNet on my dataset. after reviewing tthis problem, inspired by @renexdev, I check comleteness of my download model file and I found the model file was incomplete!!!! so the redownload the model ,finally I fix it. hoping that's helps. goog luck.
Nice @DamonDH!! Cheers
I faced the same problem and it was exactly like what @renexdev said, I replaced my current bvlc_reference_caffenet.caffemodel with the one from another computer, and it worked!
The network needs caffemodel file. In my case i was giving .solverstate file. I changed it to caffemodel file and it worked.
Most helpful comment
@ThakurRajAnand Hello, I met the same problem when I trying to finetune the ResNet on my dataset. after reviewing tthis problem, inspired by @renexdev, I check comleteness of my download model file and I found the model file was incomplete!!!! so the redownload the model ,finally I fix it. hoping that's helps. goog luck.