Comments (5)
Hey maybe i can help. You said you have created the list_attr_celeba.txt
file – did you download the
file or did you just create an empty one? The file contains quite a lot of data that's crucial for the training. You can download the txt files from my working dataset folder here: https://we.tl/fhGdPXS79t or the original one from here:
https://www.dropbox.com/sh/8oqt9vytwxb3s4r/AAA8YmAHNNU6BEfWMPMfM6r9a/Anno?dl=0
Put all the txt files in your dataset folder and try it again 😁
You also need to use the "Align&Cropped Images" archive:
https://www.dropbox.com/sh/8oqt9vytwxb3s4r/AADSNUu0bseoCKuxuI5ZeTl1a/Img?dl=0&preview=img_align_celeba.zip
not the "In-The-Wild Images".
The file should look like this:
202599
5_o_Clock_Shadow Arched_Eyebrows Attractive Bags_Under_Eyes Bald Bangs Big_Lips Big_Nose Black_Hair Blond_Hair Blurry Brown_Hair Bushy_Eyebrows Chubby Double_Chin Eyeglasses Goatee Gray_Hair Heavy_Makeup High_Cheekbones Male Mouth_Slightly_Open Mustache Narrow_Eyes No_Beard Oval_Face Pale_Skin Pointy_Nose Receding_Hairline Rosy_Cheeks Sideburns Smiling Straight_Hair Wavy_Hair Wearing_Earrings Wearing_Hat Wearing_Lipstick Wearing_Necklace Wearing_Necktie Young
000001.jpg -1 1 1 -1 -1 -1 -1 -1 -1 -1 -1 1 -1 -1 -1 -1 -1 -1 1 1 -1 1 -1 -1 1 -1 -1 1 -1 -1 -1 1 1 -1 1 -1 1 -1 -1 1
000002.jpg -1 -1 -1 1 -1 -1 -1 1 -1 -1 -1 1 -1 -1 -1 -1 -1 -1 -1 1 -1 1 -1 -1 1 -1 -1 -1 -1 -1 -1 1 -1 -1 -1 -1 -1 -1 -1 1
(... and a lot more lines 😀 )
from deep-makeover.
Just look at your errors: line 159 of dm_main.py
starts loading the training files:
train_data = _get_train_data()
line 93 sets the source images:
source_images = dm_input.input_data(sess, 'train', selected)
next, on line 11 of dm_input.py
:
assert mode == 'inference' or len(filenames) >= FLAGS.test_vectors
So maybe the filenames (and features) can't get loaded because your list_attr_celeba.txt
is empty?
from deep-makeover.
@subzerofun Hi! so I downloaded your files and put it in dataset,
Then again run it =>
69714 source images selected
78610 target images selectedGenerator input (feature) size is 100 x 80 x 3 = 24000
Generator has 0.59M parametersDiscriminator input (feature) size is 100 x 80 x 3 = 24000
Discriminator has 0.84M parametersTraceback (most recent call last):
File "dm_main.py", line 169, in
tf.app.run()
File "/Users/oteldanagul/anaconda3/lib/python3.6/site-packages/tensorflow/python/platform/app.py", line 43, in run
sys.exit(main(sys.argv[:1] + flags_passthrough))
File "dm_main.py", line 159, in main
train_data = _get_train_data()
File "dm_main.py", line 106, in _get_train_data
train_model = dm_model.create_model(sess, source_images, target_images, annealing, verbose=True)
File "/Users/oteldanagul/Downloads/deep-makeover-master/dm_model.py", line 214, in create_model
gene_loss = _generator_loss(source_images, gene_out, disc_fake_out, annealing)
File "/Users/oteldanagul/Downloads/deep-makeover-master/dm_model.py", line 135, in _generator_loss
gene_adversarial_loss = tf.nn.sigmoid_cross_entropy_with_logits(logits=disc_fake_output, labels=tf.ones_like(disc_fake_output))
TypeError: sigmoid_cross_entropy_with_logits() got an unexpected keyword argument 'labels'
from deep-makeover.
@david-gpu @subzerofun I updated version of tensorflow, then I have thats kind of results:
Am I on right way of processing images????
from deep-makeover.
David wrote in the readme:
I recommend training the model for about 40,000 to 50,000 batches.
You will need to adjust --train_time depending on how many batches/hour your system can train.
Each of these two examples were made after training a model for just two hours on one GTX 1080 GPU.
The default training time is 120min. But how far the training gets, how many batches are generated depends on the speed of your system (GPU or CPU). David uses the fastest consumer card from Nvidia out there, the GTX 1080 with 8 GB VRAM.
Does your Mac even have a Nvidia graphics card? Have you installed CUDA?
iMac or MacBook – which model?
Computation on a (recent) GPU is significantly faster than on the CPU.
Read here why:
https://www.quora.com/Why-are-GPUs-well-suited-to-deep-learning
http://navoshta.com/cpu-vs-gpu/
So if you don't have a GPU (or haven't configured it) Tensorflow will use your CPU.
My Nvidia GTX 780 generated 16 batch pngs in 25min (batch nr. 3000).
So if you let it run on your machine for two hours and only got 3 pngs, then your MacBook/iMac is simply too slow! As David said, you need 40,000 to 50,000 batches to get good results. If you got 3 pngs in 120min, then you only have trained 400 batches. That's not enough to generate believable images!
This code is intended for training on the GPU – or on a powerful multicore processor.
Check if your Mac has a graphics card – then try to install CUDA (for Nvidia) or OpenCL (for AMD) and configure it for Tensorflow.
https://www.tensorflow.org/get_started/os_setup
from deep-makeover.
Related Issues (13)
- Asap I can't make my boss more pretty. ))) HOT 3
- Results not repeatable? HOT 4
- How to get the key in dm_input.py?? HOT 2
- Getting incorrect images as output
- Is there a paper?
- ResourceExhaustedError (see above for traceback): OOM when allocating tensor with shape[16,100,80,48] HOT 1
- When training for longer time (>7 hours) fails to save checkpoint HOT 5
- Longer training and higher resolution images? HOT 2
- AttributeError: module 'tensorflow' has no attribute 'Variable'
- Where I can get Large-scale CelebFaces Attributes (CelebA) Dataset ??? HOT 4
- How much should be done training? HOT 9
- AssertionError
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from deep-makeover.