Closed yja1 closed 7 years ago
How did you solve the issue? @yja1
How did you solve the issue? @yja1
在同目录下做一个libcaffe的软连接
------------------ 原始邮件 ------------------ 发件人: "Ashutosh Mishra";notifications@github.com; 发送时间: 2017年7月26日(星期三) 下午4:19 收件人: "rksltnl/Deep-Metric-Learning-CVPR16"Deep-Metric-Learning-CVPR16@noreply.github.com; 抄送: "1094110236"1094110236@qq.com; "Mention"mention@noreply.github.com; 主题: Re: [rksltnl/Deep-Metric-Learning-CVPR16]gen_caffe_dataset_multilabel_m128.m run error (#15)
How did you solve the issue? @yja1
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.
Can you please write it in english?
make a link for libcaffe.so in the same folder with gen_caffe_dataset_multilabel_m128.m
------------------ 原始邮件 ------------------ 发件人: "Ashutosh Mishra";notifications@github.com; 发送时间: 2017年7月26日(星期三) 下午5:16 收件人: "rksltnl/Deep-Metric-Learning-CVPR16"Deep-Metric-Learning-CVPR16@noreply.github.com; 抄送: "1094110236"1094110236@qq.com; "Mention"mention@noreply.github.com; 主题: Re: [rksltnl/Deep-Metric-Learning-CVPR16]gen_caffe_dataset_multilabel_m128.m run error (#15)
Can you please write it in english?
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.
Libcaffe.so is in the caffe directory i.e. /caffe/build/lib and the gen_caffe_dataset_multilabel_m128.m matlab code is in code directory of the project. Could you please elaborate more as to how to make the link?
or you can copy libcaffe.so.1.0.0-rc3 (in the caffe directory) to gen_caffe_dataset_multilabel_m128.m folder
------------------ 原始邮件 ------------------ 发件人: "Ashutosh Mishra";notifications@github.com; 发送时间: 2017年7月26日(星期三) 下午5:30 收件人: "rksltnl/Deep-Metric-Learning-CVPR16"Deep-Metric-Learning-CVPR16@noreply.github.com; 抄送: "1094110236"1094110236@qq.com; "Mention"mention@noreply.github.com; 主题: Re: [rksltnl/Deep-Metric-Learning-CVPR16]gen_caffe_dataset_multilabel_m128.m run error (#15)
Libcaffe.so is in the caffe directory i.e. /caffe/build/lib and the gen_caffe_dataset_multilabel_m128.m code is in code directory of the project. Could you please elaborate more as to how to make the link?
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.
I copied libcaffe.so in the folder containing the gen_caffe_dataset_multilabel_m128.m code but it gives the following error:
Invalid MEX-file '/users/amishra/code/serialized_pairs_to_leveldb.mexa64': /users/amishra/caffe/caffe-master/build/lib/libcaffe.so.1.0.0: undefined symbol:_ZN2cv8imencodeERKNSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEERKNS_11_InputArrayERSt6vectorIhSaIhEERKSB_IiSaIiEE
copy libcaffe.so.1.0.0-rc3 not libcaffe.so. and run in .m file folder: ln -s libcaffe.so.1.0.0-rc3 libcaffe.so
------------------ 原始邮件 ------------------ 发件人: "Ashutosh Mishra";notifications@github.com; 发送时间: 2017年7月26日(星期三) 晚上6:02 收件人: "rksltnl/Deep-Metric-Learning-CVPR16"Deep-Metric-Learning-CVPR16@noreply.github.com; 抄送: "1094110236"1094110236@qq.com; "Mention"mention@noreply.github.com; 主题: Re: [rksltnl/Deep-Metric-Learning-CVPR16]gen_caffe_dataset_multilabel_m128.m run error (#15)
I copied libcaffe.so in the folder containing the gen_caffe_dataset_multilabel_m128.m code but it gives the following error:
Invalid MEX-file '/users/amishra/code/serialized_pairs_to_leveldb.mexa64': /users/amishra/caffe/caffe-master/build/lib/libcaffe.so.1.0.0: undefined symbol:_ZN2cv8imencodeERKNSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEERKNS_11_InputArrayERSt6vectorIhSaIhEERKSB_IiSaIiEE
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.
I did but after executing the command libcaffe.so pops up and I get the same error as before.
you can try this, in .m file folder run:
rm libcaffe.so
ln -s libcaffe.so.1.0.0-rc3 /users/amishra/caffe/caffe-master/build/lib/libcaffe.so
ln -s libcaffe.so.1.0.0-rc3 /users/amishra/caffe/caffe-master/build/lib/libcaffe.so.1.0.0
------------------ 原始邮件 ------------------ 发件人: "Ashutosh Mishra";notifications@github.com; 发送时间: 2017年7月26日(星期三) 晚上7:04 收件人: "rksltnl/Deep-Metric-Learning-CVPR16"Deep-Metric-Learning-CVPR16@noreply.github.com; 抄送: "1094110236"1094110236@qq.com; "Mention"mention@noreply.github.com; 主题: Re: [rksltnl/Deep-Metric-Learning-CVPR16]gen_caffe_dataset_multilabel_m128.m run error (#15)
I did but after executing the command libcaffe.so pops up and I get the same error as before.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.
I did it and it says:
ln: failed to create symbolic link '/users/amishra/caffe/caffe-master/build/lib/libcaffe.so.1.0.0': File exists
can you show you folder files of gen_caffe_dataset_multilabel_m128.m
------------------ 原始邮件 ------------------ 发件人: "Ashutosh Mishra";notifications@github.com; 发送时间: 2017年7月26日(星期三) 晚上7:14 收件人: "rksltnl/Deep-Metric-Learning-CVPR16"Deep-Metric-Learning-CVPR16@noreply.github.com; 抄送: "1094110236"1094110236@qq.com; "Mention"mention@noreply.github.com; 主题: Re: [rksltnl/Deep-Metric-Learning-CVPR16]gen_caffe_dataset_multilabel_m128.m run error (#15)
I did it and it says:
ln: failed to create symbolic link '/users/amishra/caffe/caffe-master/build/lib/libcaffe.so.1.0.0': File exists
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.
These are the files in the folder containing the matlab code of gen_caffe_multilabel_m128.m
128.py config.m evaluation find_dups.m gen_caffe_dataset_multilabel_m128.m gen_caffe_validation_imageset.m gen_images.m gen_splits.m get_training_examples_multilabel.m IIITCFWdata.mat imageset_to_leveldb.cpp imageset_to_leveldb.mexa64 libcaffe.so.1.0.0-rc3 load_cropped_images.m queryInfo.mat realFaces savefast.m serialized_pairs_to_leveldb.cpp serialized_pairs_to_leveldb.mexa64 serialized_triplets_to_leveldb.cpp splits.mat training_images_crop15_square256.mat validation_images_crop15_square256.mat
sorry, run this:
rm libcaffe.so
ln -s /users/amishra/caffe/caffe-master/build/lib/libcaffe.so.1.0.0-rc3 libcaffe.so
ln -s /users/amishra/caffe/caffe-master/build/lib/libcaffe.so.1.0.0-rc3 libcaffe.so.1.0.0
------------------ 原始邮件 ------------------ 发件人: "Ashutosh Mishra";notifications@github.com; 发送时间: 2017年7月26日(星期三) 晚上7:25 收件人: "rksltnl/Deep-Metric-Learning-CVPR16"Deep-Metric-Learning-CVPR16@noreply.github.com; 抄送: "1094110236"1094110236@qq.com; "Mention"mention@noreply.github.com; 主题: Re: [rksltnl/Deep-Metric-Learning-CVPR16]gen_caffe_dataset_multilabel_m128.m run error (#15)
These are the files in the folder containing the matlab code of gen_caffe_multilabel_m128.m
128.py config.m evaluation find_dups.m gen_caffe_dataset_multilabel_m128.m gen_caffe_validation_imageset.m gen_images.m gen_splits.m get_training_examples_multilabel.m IIITCFWdata.mat imageset_to_leveldb.cpp imageset_to_leveldb.mexa64 libcaffe.so.1.0.0-rc3 load_cropped_images.m queryInfo.mat realFaces savefast.m serialized_pairs_to_leveldb.cpp serialized_pairs_to_leveldb.mexa64 serialized_triplets_to_leveldb.cpp splits.mat training_images_crop15_square256.mat validation_images_crop15_square256.mat
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.
I did exactly what you said but the error still not goes, rather now it is:
Invalid MEX-file '/users/amishra/code/serialized_pairs_to_leveldb.mexa64': libcaffe.so.1.0.0: cannot open shared object file: No such file or directory
Hey @yja1 @rksltnl , please suggest some alternative to this error
The reason it gives error I think because both the .so files are in the same folder.
Do they need to be in the same or what needs to be done?
Did you follow the Caffe installation instructions? Are Pycaffe and Matcaffe installed?
Yes I have installed caffe according to the instructions given on the CAFFE page
Which version of Caffe? Did you install the included submodule version? https://github.com/rksltnl/Caffe-Deep-Metric-Learning-CVPR16
No, I didn't install the submodule version. I am following the steps of training in the first part. Do I need to install it?
Yes it's possible that something got changed in recent Caffe and isn't compatible with the submodule version. Please try the submodule Caffe version.
I figured out that matcaffe wasn't installed and it was a mistake on my part. After doing the necessary installations and re-checking it again, though I did not install the Caffe submodule version as it is giving some error to build, I get the same error Invalid MEX file error when following the basic installation steps
Please help me in this !!
With which version of cuDNN have you compiled the submodule version as I have CUDA version 8 and cuDNN version 5.1.5 but it gives error in cudnn.hpp: error: argument of type "int" is incompatible with parameter of type "cudnnNanPropagation_t"
执行gen_caffe_dataset_multilabel_m128.m 的时候报错: Deleted 31 data pairs from negatives for batch divisibility Writing level db.. Invalid MEX-file '/home/yja/caffecode/lifted_caffe/code/serialized_pairs_to_leveldb.mexa64': libcaffe.so.1.0.0-rc3: cannot open shared object file: No such file or directory Error in gen_caffe_dataset_multilabel_m128 (line 88) 但是已经做了compile.m 生成了serialized_pairs_to_leveldb.mexa64。希望您帮看下,谢谢!