Dataset

[xyzdcgan]

Hello can you help me with the dataset when i go to dataset link there i download img_celeba.7z it download files but which are not jpg files and there are files like img_celeba.7z.001 so from where we can download the images. there are 3 folder img_align_celeba.zip which are jpg files but with 178-218 resolution. so where are the high resolution image...??

[LynnHo]

@xyzdcgan Download files in this folder and unzip the files.

[xyzdcgan]

@LynnHo Okay, from that 3 are getting download, i am using drive link, so once it complete zipping that it start download of 3 files which are and also extracted those and got these result like:

[LynnHo]

@xyzdcgan

There should be 14 files like above, you can use 7z x img_celeba.7z.001 to extract them.

[xyzdcgan]

@LynnHo That gives Error : E_FAIL
7z x img_celeba.7z.001

7-Zip [64] 9.20 Copyright (c) 1999-2010 Igor Pavlov 2010-11-18
p7zip Version 9.20 (locale=en_IN,Utf16=on,HugeFiles=on,4 CPUs)

Processing archive: img_celeba.7z.001

Error: E_FAIL

Trying to solve this if you can have any solution for this it would be great

I have plenty of space so why does,

Do i need to download all 14 parts

[LynnHo]

@xyzdcgan Yes, you need all 14 parts.

[xyzdcgan]

@LynnHo Okay

@xyzdcgan Yes, you need all 14 parts.

Okay

[xyzdcgan]

@LynnHo Also i am trying to convert the checkpoint files to .pb which is done through tensorflow frozen_graph.py in which i pass Adam And adam_1 as the output nodes and which is generated but when i try it out it gives and error of adam not convert with float32 to float32_ref. can you tell if you have used any other optimizer to train the model as it only accept adam as output node. not relu or other

[LynnHo]

@xyzdcgan I am sorry that I haven't got the experience on this problem.

@LynnHo Also i am trying to convert the checkpoint files to .pb which is done through tensorflow frozen_graph.py in which i pass Adam And adam_1 as the output nodes and which is generated but when i try it out it gives and error of adam not convert with float32 to float32_ref. can you tell if you have used any other optimizer to train the model as it only accept adam as output node. not relu or other

[xyzdcgan]

@LynnHo Have you ever converted your ckpt files to .pb extension file format.

[LynnHo]

@xyzdcgan No, I haven't.

[xyzdcgan]

@LynnHo

python3 hd_celeba.py --data_dir ./data --crop_size 512 --order 4 --save_format jpg --n_worker 32
Use OpenCV
0%| | 0/202599 [00:00<?, ?it/s]multiprocessing.pool.RemoteTraceback:
"""
Traceback (most recent call last):
File "/usr/lib/python3.6/multiprocessing/pool.py", line 119, in worker
result = (True, func(*args, **kwds))
File "hd_celeba.py", line 105, in work
mode=args.mode)
File "/home/ios/hdimage/HD-CelebA-Cropper-master/cropper.py", line 64, in align_crop_5pts_opencv
move = np.array([img.shape[1] // 2, img.shape[0] // 2])
AttributeError: 'NoneType' object has no attribute 'shape'
"""

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
File "hd_celeba.py", line 109, in
for _ in tqdm(pool.imap(work, range(len(img_names))), total=len(img_names)):
File "/home/ios/.local/lib/python3.6/site-packages/tqdm/_tqdm.py", line 999, in iter
for obj in iterable:
File "/usr/lib/python3.6/multiprocessing/pool.py", line 761, in next
raise value
AttributeError: 'NoneType' object has no attribute 'shape'

[LynnHo]

@xyzdcgan Did you follow the readme step by step? It seems to be the problem of image loading, how did you organize the files?

[LynnHo]

@xyzdcgan try python3 hd_celeba.py --data_dir . --crop_size 512 --order 4 --save_format jpg --n_worker 32

[xyzdcgan]

@xyzdcgan try python3 hd_celeba.py --data_dir . --crop_size 512 --order 4 --save_format jpg --n_worker 32

Working now Thanks

[copperwiring]

@xyzdcgan Were you able to get the data (.7z files) from the 'images' directory in CelebA? To me it doens't work and throws cannot open the file as archive error