Closed Kelly-ZH closed 3 years ago
@Kelly-ZH
Hmm... It works on my Ubuntu machine, so I guess it's a Windows problem. I'm sorry, but I don't have a Windows environment, so I don't think I can help.
@Kelly-ZH
What is the file size of the MPIIFaceGaze.h5
created by tools/preprocess_mpiifacegaze.py
? It should be about 26GB.
And if so, what's the result of the following command?
f = h5py.File('MPIIFaceGaze.h5', 'r')
print(f.keys())
@Kelly-ZH
What is the file size of the
MPIIFaceGaze.h5
created bytools/preprocess_mpiifacegaze.py
? It should be about 26GB. And if so, what's the result of the following command?f = h5py.File('MPIIFaceGaze.h5', 'r') print(f.keys())
The size of my MPIIFaceGaze.h5 is 10.1 GB, MPIIFaceGaze_normalized is 21.2GB, maybe I should run tools/preprocess_mpiifacegaze.py again. And the result of the following command
is<KeysViewHDF5 ['p00']>
@Kelly-ZH
The result of the aforementioned command should look like the following:
<KeysViewHDF5 ['p00', 'p01', 'p02', 'p03', 'p04', 'p05', 'p06', 'p07', 'p08', 'p09', 'p10', 'p11', 'p12', 'p13', 'p14']>
It seems the dataset you created only contains p00
.
Try deleting it and creating it again.
@Kelly-ZH
The result of the aforementioned command should look like the following:
<KeysViewHDF5 ['p00', 'p01', 'p02', 'p03', 'p04', 'p05', 'p06', 'p07', 'p08', 'p09', 'p10', 'p11', 'p12', 'p13', 'p14']>
It seems the dataset you created only contains
p00
. Try deleting it and creating it again.
Oh ,I will check it. And the size of my MPIIFaceGaze_normalized.zip is 20.7 GB, is it true? If not, could you share me the MPIIFaceGaze_normalized.zip? It will takes several days for me to download the zip because of my terrible network. Thank you very much.
@Kelly-ZH
I think the size of your MPIIFaceGaze_normalized.zip
is as expected. FYI, its md5sum is df2182d207a57818ebf17460dd9014d8
, so you can check if it's corrupted.
@Kelly-ZH
I think the size of your
MPIIFaceGaze_normalized.zip
is as expected. FYI, its md5sum isdf2182d207a57818ebf17460dd9014d8
, so you can check if it's corrupted.
I run
python tools/preprocess_mpiifacegaze.py --dataset datasets/MPIIFaceGaze_normalized -o datasets/ then an error has occured, that is
73%|█████████████████████████████████████████████████▊ | 11/15 [2:48:13<1:01:10, 917.59s/it] Traceback (most recent call last): File "D:\project\Code\pytorch_mpiigaze-master\tools\preprocess_mpiifacegaze.py", line 51, in
main() File "D:\project\Code\pytorch_mpiigaze-master\tools\preprocess_mpiifacegaze.py", line 47, in main add_mat_data_to_hdf5(person_id, dataset_dir, output_path) File "D:\project\Code\pytorch_mpiigaze-master\tools\preprocess_mpiifacegaze.py", line 18, in add_mat_datato hdf5 images = images.transpose(0, 2, 3, 1).astype(np.uint8) MemoryError: Unable to allocate 1.68 GiB for an array with shape (3000, 448, 448, 3) and data type uint8 What's wrong with it? Thank you.
@Kelly-ZH
As you can see in the error message, I think your machine doesn't have enough memory.
Hi hysts, Thank you for your work very much, And when I run "python train.py --config configs/mpiifacegaze/alexnet_train.yaml", there was this mistake like this: