problems about h5 file list

[bluesliuf]

谢谢你分享你的代码，我按照你的步骤尝试复现模型，运行generate_h5_300w_scale_V4.py 扩充数据集并且生成h5文件和列表，我想问一下，生成的train.txt和valid.txt列表文件里面只有1行吗？类似这样： h5_300w_68_40_box/train.h5149755 h5_300w_68_40_box/valid.h5127920 我运行train.sh后 报错

I0719 17:53:03.728138 3599 layer_factory.hpp:77] Creating layer I0719 17:53:03.728199 3599 net.cpp:86] Creating Layer I0719 17:53:03.728371 3599 net.cpp:382] -> data I0719 17:53:03.728436 3599 net.cpp:382] -> landmarks I0719 17:53:03.728488 3599 hdf5_data_layer.cpp:81] Loading list of HDF5 filenames from: /home/boyun/data/liufan/landmark/code/Vanilla_CNN/h5_300w_68_40_box/train.txt I0719 17:53:03.728675 3599 hdf5_data_layer.cpp:95] Number of HDF5 files: 1 HDF5-DIAG: Error detected in HDF5 (1.8.16) thread 139798604674880: #000: ../../../src/H5F.c line 604 in H5Fopen(): unable to open file major: File accessibilty minor: Unable to open file #001: ../../../src/H5Fint.c line 992 in H5F_open(): unable to open file: time = Fri Jul 19 17:53:03 2019 , name = 'h5_300w_68_40_box/train.h5149755', tent_flags = 0 major: File accessibilty minor: Unable to open file #002: ../../../src/H5FD.c line 993 in H5FD_open(): open failed major: Virtual File Layer minor: Unable to initialize object #003: ../../../src/H5FDsec2.c line 339 in H5FD_sec2_open(): unable to open file: name = 'h5_300w_68_40_box/train.h5149755', errno = 2, error message = 'No such file or directory', flags = 0, o_flags = 0 major: File accessibilty minor: Unable to open file F0719 17:53:03.733265 3599 hdf5_data_layer.cpp:32] Failed opening HDF5 file: h5_300w_68_40_box/train.h5149755 *** Check failure stack trace: *** @ 0x7f2564dfa5cd google::LogMessage::Fail() @ 0x7f2564dfc433 google::LogMessage::SendToLog() @ 0x7f2564dfa15b google::LogMessage::Flush() @ 0x7f2564dfce1e google::LogMessageFatal::~LogMessageFatal() @ 0x7f25655556b3 caffe::HDF5DataLayer<>::LoadHDF5FileData() @ 0x7f2565552417 caffe::HDF5DataLayer<>::LayerSetUp() @ 0x7f2565575437 caffe::Net<>::Init() @ 0x7f2565577b7e caffe::Net<>::Net() @ 0x7f25655e75fa caffe::Solver<>::InitTrainNet() @ 0x7f25655e8ac5 caffe::Solver<>::Init() @ 0x7f25655e8ddf caffe::Solver<>::Solver() @ 0x7f25655c7c61 caffe::Creator_AdamSolver<>() @ 0x40a758 train() @ 0x407548 main @ 0x7f2563590830 __libc_start_main @ 0x407e19 _start @ (nil) (unknown) 是不是生成的列表有问题，希望您能够帮忙解答一下

[cunjian]

I believe you should remove the number of samples. h5_300w_68_40_box/train.h5 h5_300w_68_40_box/valid.h5

[bluesliuf]

谢谢你的回复，这个问题之前在我提问后已经解决了。还想请教您，如果实现celeba 5点数据集，数据量够大了，不做数据增强的话，是不是需要将generate_h5py()方法中的augment设置为false。还有作者是不是没有提供Tweaked 部分的代码，只提供了Vanilla CNN的代码呀

[cunjian]

谢谢你的回复，这个问题之前在我提问后已经解决了。还想请教您，如果实现celeba 5点数据集，数据量够大了，不做数据增强的话，是不是需要将generate_h5py()方法中的augment设置为false。还有作者是不是没有提供Tweaked 部分的代码，只提供了Vanilla CNN的代码呀

Not sure about the original implementation. You can check their details.