Week5- CycleGAN试用感受
https://github.com/vanhuyz/CycleGAN-TensorFlow
First, download a dataset, e.g. apple2orange
$ bash download_dataset.sh apple2orange
http://ju.outofmemory.cn/entry/326146
https://github.com/yuxiaofelicia/CycleGAN-TensorFlow
https://github.com/tensorflow/models/blob/master/research/slim/nets/cyclegan.py
https://mlnotebook.github.io/post/GAN4/#buildmodel
我倒是下载了它的数据集
深度学习数据集好大
下载都下载好久
https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix
这个是原著作者的
命令行下载数据集的时候也下载好久。
我可找不到这么多数据集来train人家
我可能只能用pretained模型来搞我自己的应用
伯克利的原作者也上传了大量的训练好的模型了
1. 下载项目代码
git clone https://github.com/vanhuyz/CycleGAN-TensorFlow.git
2. 将图片转换成tfrecords格式
这个项目中提供了一个build_data脚本，用于将图片转换成tfrecords形式。假设我们的图片存放在~/datasets/man2woman/a_resized/和 ~/datasets/man2woman/b_resized目录下，对应的命令就是：
python3 build_data.py \
--X_input_dir ~/Downloads/man2woman/a_resized/ \
--Y_input_dir ~/Downloads/man2woman/b_resized/ \
--X_output_file ~/Downloads/man2woman/man.tfrecords \
--Y_output_file ~/Downloads/man2woman/woman.tfrecords
3. 训练
训练的命令为：
python train.py \
--X ~/datasets/man2woman/man.tfrecords \
--Y ~/datasets/man2woman/woman.tfrecords \
--image_size 256
训练的过程比较漫长，此时可以打开TensorBoard来观察训练情况（运行这个命令时需要将“20170715-1622”改成机器中对应的文件夹，下同）：
tensorboard --logdir checkpoints/20170715-1622
4. 导出模型并执行单张图片
导出模型的方法为：
python export_graph.py \
--checkpoint_dir checkpoints/20170715-1622 \
--XtoY_model man2woman.pb \
--YtoX_model woman2man.pb \
--image_size 256
对单张图片进行转换（将data/test.jpg替换为对应的输入图片地址）：
python inference.py \
--model pretrained/man2woman.pb \
--input data/test.jpg \
--output data/output.jpg \
--image_size 256
Felicias-MacBook-Air:~ felicia$ cd CycleGAN-TensorFlow
Felicias-MacBook-Air:CycleGAN-TensorFlow felicia$ python build_data.py \
> --X_input_dir ~/Downloads/man2woman/a_resized/ \
> --Y_input_dir ~/Downloads/man2woman/b_resized/ \
> --X_output_file ~/Downloads/man2woman/man.tfrecords \
> --Y_output_file ~/Downloads/man2woman/woman.tfrecords
Traceback (most recent call last):
File "/Users/felicia/anaconda3/lib/python3.6/site-packages/tensorflow/python/pywrap_tensorflow.py", line 41, infrom tensorflow.python.pywrap_tensorflow_internal import * File "/Users/felicia/anaconda3/lib/python3.6/site-packages/tensorflow/python/pywrap_tensorflow_internal.py", line 28, in_pywrap_tensorflow_internal = swig_import_helper() File "/Users/felicia/anaconda3/lib/python3.6/site-packages/tensorflow/python/pywrap_tensorflow_internal.py", line 24, in swig_import_helper _mod = imp.load_module('_pywrap_tensorflow_internal', fp, pathname, description) File "/Users/felicia/anaconda3/lib/python3.6/imp.py", line 243, in load_module return load_dynamic(name, filename, file) File "/Users/felicia/anaconda3/lib/python3.6/imp.py", line 343, in load_dynamic return _load(spec)ImportError: dlopen(/Users/felicia/anaconda3/lib/python3.6/site-packages/tensorflow/python/_pywrap_tensorflow_internal.so, 10): Library not loaded: @rpath/libcublas.8.0.dylib Referenced from: /Users/felicia/anaconda3/lib/python3.6/site-packages/tensorflow/python/_pywrap_tensorflow_internal.so Reason: image not foundDuring handling of the above exception, another exception occurred:Traceback (most recent call last): File "build_data.py", line 1, inimport tensorflow as tf File "/Users/felicia/anaconda3/lib/python3.6/site-packages/tensorflow/__init__.py", line 24, infrom tensorflow.python import * File "/Users/felicia/anaconda3/lib/python3.6/site-packages/tensorflow/python/__init__.py", line 51, infrom tensorflow.python import pywrap_tensorflow File "/Users/felicia/anaconda3/lib/python3.6/site-packages/tensorflow/python/pywrap_tensorflow.py", line 52, inraise ImportError(msg)ImportError: Traceback (most recent call last): File "/Users/felicia/anaconda3/lib/python3.6/site-packages/tensorflow/python/pywrap_tensorflow.py", line 41, infrom tensorflow.python.pywrap_tensorflow_internal import * File "/Users/felicia/anaconda3/lib/python3.6/site-packages/tensorflow/python/pywrap_tensorflow_internal.py", line 28, in _pywrap_tensorflow_internal = swig_import_helper()
File "/Users/felicia/anaconda3/lib/python3.6/site-packages/tensorflow/python/pywrap_tensorflow_internal.py", line 24, in swig_import_helper
_mod = imp.load_module('_pywrap_tensorflow_internal', fp, pathname, description)
File "/Users/felicia/anaconda3/lib/python3.6/imp.py", line 243, in load_module
return load_dynamic(name, filename, file)
File "/Users/felicia/anaconda3/lib/python3.6/imp.py", line 343, in load_dynamic
return _load(spec)
ImportError: dlopen(/Users/felicia/anaconda3/lib/python3.6/site-packages/tensorflow/python/_pywrap_tensorflow_internal.so, 10): Library not loaded: @rpath/libcublas.8.0.dylib
Referenced from: /Users/felicia/anaconda3/lib/python3.6/site-packages/tensorflow/python/_pywrap_tensorflow_internal.so
Reason: image not found
Failed to load the native TensorFlow runtime.
See https://www.tensorflow.org/install/install_sources#common_installation_problems
for some common reasons and solutions. Include the entire stack trace
above this error message when asking for help.
好像是因为装了1.1和1.5
卸载了1.5就可以顺利import了
没办法 不
DATA_ROOT=./datasets/ae_photos
name=style_cezanne_pretrained
model=one_direction_test
phase=test
loadSize=256
fineSize=256
resize_or_crop="scale_width"
th test.lua
