euphoriayan / zi2zi-pytorch Goto Github PK

Is it because they are not effective as much as original repo suggested?

Many thanks for your work! I think the font2img.py is better thank the original repo.

Try to use 512512 , can't train. is it possible to change the image size ? as my test , the only one which can be put into training is 256256 .

thanks!

作者您好！为何在最后一步生成的图片中文字显示不完整，请问该如何解决？谢谢！

您好 我使用自己拼接的图片做的数据集，并没有从ttf中提取，请问infer的时候如何输入一张我的汉字图片，去测试效果呢，谢谢！

预训练模型链接过期了能补一下吗？感谢

推理阶段，希望生成自己手写的汉字，那以下命令中src_font 怎么赋值呢？
python infer.py --experiment_dir experiment
--gpu_ids cuda:0
--batch_size 32
--resume {the saved model you select}
--from_txt
--src_font {your model's source font file}
--src_txt 大威天龙大罗法咒世尊地藏波若诸佛
--label 3

/pytorch/aten/src/ATen/native/cuda/Indexing.cu:609: indexSelectSmallIndex: block: [0,0,0], thread: [122,0,0] Assertion srcIndex < srcSelectDimSize failed.
/pytorch/aten/src/ATen/native/cuda/Indexing.cu:609: indexSelectSmallIndex: block: [0,0,0], thread: [123,0,0] Assertion srcIndex < srcSelectDimSize failed.
/pytorch/aten/src/ATen/native/cuda/Indexing.cu:609: indexSelectSmallIndex: block: [0,0,0], thread: [124,0,0] Assertion srcIndex < srcSelectDimSize failed.
/pytorch/aten/src/ATen/native/cuda/Indexing.cu:609: indexSelectSmallIndex: block: [0,0,0], thread: [125,0,0] Assertion srcIndex < srcSelectDimSize failed.
/pytorch/aten/src/ATen/native/cuda/Indexing.cu:609: indexSelectSmallIndex: block: [0,0,0], thread: [126,0,0] Assertion srcIndex < srcSelectDimSize failed.
/pytorch/aten/src/ATen/native/cuda/Indexing.cu:609: indexSelectSmallIndex: block: [0,0,0], thread: [127,0,0] Assertion srcIndex < srcSelectDimSize failed.
......
Traceback (most recent call last):
File "infer.py", line 154, in
main()
File "infer.py", line 144, in main
model.sample(batch, infer_dir)
result = self.forward(*input, **kwargs)
File "/home/pytorch/lib/python3.6/site-packages/torch/nn/modules/conv.py", line 399, in forward
return self._conv_forward(input, self.weight, self.bias)
File "/home/lib/python3.6/site-packages/torch/nn/modules/conv.py", line 396, in _conv_forward
self.padding, self.dilation, self.groups)
RuntimeError: cuDNN error: CUDNN_STATUS_INTERNAL_ERROR

conda env export > environment.yaml // 导出当前环境的包信息

conda env create -f environment.yaml // 用配置文件创建新的虚拟环境

pip freeze > piplist.txt pip freeze > piplist.yaml // 导出pip安装的所有的包

pip install -r piplist.txt // 在新的环境中安装导出的包

hello,img2img,
ch = res[1]
TypeError: 'NoneType' object is not subscriptable

RuntimeError: output with shape [1, 256, 256] doesn't match the broadcast shape [3, 256, 256]

求教下，怎么调整？

第一步就报错'NoneType' object has no attribute 'tobytes'
是因为包的版本不匹配嘛

执行
ython train.py --experiment_dir=experiment --gpu_ids=cuda:0 --batch_size=32 --epoch=100 --sample_steps=200 --checkpoint_steps=500

的时候 遇到这个问题

一直提示这个，明明按照你的，加了--mode

python font2img.py --src_font='source.ttf'
--dst_font='cao.ttf'
--charset=CN
--sample_count=1000
--sample_dir='images'
--label=0
--filter
--shuffle
--mode=font2font
在使用如上命令的时候，应该是font2img.py文件里的这句：
img = draw_single_char(c, font, canvas_size, x_offset, y_offset)
出错了，求解决方法

unpickled total 0 examples
unpickled total 8517 examples
Traceback (most recent call last):
File "train.py", line 128, in
main()
File "train.py", line 102, in main
for bid, batch in enumerate(dataloader):
File "D:\PaddleGAN-venv\lib\site-packages\torch\utils\data\dataloader.py", line 521, in next
data = self._next_data()
File "D:\PaddleGAN-venv\lib\site-packages\torch\utils\data\dataloader.py", line 561, in _next_data
data = self._dataset_fetcher.fetch(index) # may raise StopIteration
File "D:\PaddleGAN-venv\lib\site-packages\torch\utils\data_utils\fetch.py", line 49, in fetch
data = [self.dataset[idx] for idx in possibly_batched_index]
File "D:\PaddleGAN-venv\lib\site-packages\torch\utils\data_utils\fetch.py", line 49, in
data = [self.dataset[idx] for idx in possibly_batched_index]
File "D:\code\zi2zi-pytorch-master\data\dataset.py", line 35, in getitem
img_A, img_B = self.process(item[1])
File "D:\code\zi2zi-pytorch-master\data\dataset.py", line 102, in process
img_A = self.transform(img_A)
File "D:\PaddleGAN-venv\lib\site-packages\torch\nn\modules\module.py", line 1102, in _call_impl
return forward_call(*input, **kwargs)
File "D:\PaddleGAN-venv\lib\site-packages\torchvision\transforms\transforms.py", line 226, in forward

return F.normalize(tensor, self.mean, self.std, self.inplace)

File "D:\PaddleGAN-venv\lib\site-packages\torchvision\transforms\functional.py", line 351, in normali
ze
tensor.sub_(mean).div_(std)
RuntimeError: output with shape [1, 256, 256] doesn't match the broadcast shape [3, 256, 256]

求教下，怎么调整？

1. How is this an improvement compared to the other base libraries?
2. Would it be good to increase model variety with more fonts? kaonashi-tyc/zi2zi#74

n our pre-trained model, the mapping relationships between label and writers are below:

package.py与raw_package.py有什么区别

package.py中的parser.add_argument('--dst_json', type=str, default=None) 的dst_json文件，指的是项目下哪个文件呢

How can I get a copy of the handwriting dataset as shown in the pre-trained data?

Also, may I ask about the training details? How long does it take to get the performance similar to the pre-trained model? And what is the specs? Thanks a lot.

正常么

Hello, I follow your steps to resume the work. But I can not generate imgs. Can you help me?

谢谢您出色的工作，我需要和zi2zi做对比实验，数据集为28种字体，大小为80*80，我在上采样加了一层转置卷积使得程序能够运行起来，卷积层组为6组（之前为8组），但是infer阶段加载模型时加载鉴别器出现了错误，只加载生成器没有错误，错误如下：
RuntimeError: Error(s) in loading state_dict for Discriminator:
size mismatch for binary.weight: copying a param with shape torch.Size([1, 100]) from checkpoint, the shape in current model is torch.Size([1, 1024]).
size mismatch for catagory.weight: copying a param with shape torch.Size([40, 100]) from checkpoint, the shape in current model is torch.Size([40, 1024]).

我在使用 Google Colab 跑 zi2zi-pytorch，中断了以后如何从 checkpoint 文件恢复进度呢？

如题，如何存储为新的ttf字体？

Excuse me,have you met log like this ? It's really an annoying problem.I have set up required environment, but it didn't work.

您好，我尝试您给的模型生成，但是结果似乎都是毛笔风格，如果我给的是瘦体/或者是钢笔风格的话，生成很奇怪。 并且请问可以一次输入几个文字一起生成吗？

按照视频教程一步一步来，软件版本都降级了，但还是出现错误，请大佬解惑！
Traceback (most recent call last):
File "font2img.py", line 340, in
args.sample_count, args.sample_dir, args.label, args.filter)
File "font2img.py", line 143, in font2font
filter_hashes = set(filter_recurring_hash(charset, dst_font, canvas_size, x_offset, y_offset))
File "font2img.py", line 131, in filter_recurring_hash
hash_count[hash(img.tobytes())] += 1
AttributeError: 'NoneType' object has no attribute 'tobytes'

我尝试下载下来解压不了，显示文件出错了，其他的解压工具显示zip为空，不知道您能否给我发一份代码，我的邮箱是[email protected],万分感谢

更改infer.py中的parser.add_argument('--input_nc', type=int, default=3)后在generator.py中报如下错误：
RuntimeError: Given groups=1, weight of size [64, 3, 4, 4], expected input[13, 1, 256, 256] to have 3 channels, but got 1 channels instead

(pytorch) miao@miao:~/data/file/bwt/zi2zi-pytorch-master$ python font2img.py --src_font=src.ttf
usage: font2img.py [-h] --mode {imgs2imgs,font2imgs,font2font,fonts2imgs}
[--src_font SRC_FONT] [--src_fonts_dir SRC_FONTS_DIR]
[--src_imgs SRC_IMGS] [--dst_font DST_FONT]
[--dst_imgs DST_IMGS] [--filter] [--charset CHARSET]
[--shuffle] [--char_size CHAR_SIZE]
[--canvas_size CANVAS_SIZE] [--x_offset X_OFFSET]
[--y_offset Y_OFFSET] [--sample_count SAMPLE_COUNT]
[--sample_dir SAMPLE_DIR] [--label LABEL]
font2img.py: error: the following arguments are required: --mode

请问
font2imgs 的src_font dst_imgs sample_dir

Imgs2Imgs的 src_imgs sample_dir

分别都是什么样的数据
font2font都运行完成了 a-b a-c的

zi2zi: Master Chinese Calligraphy with Conditional Adversarial Networks
请问这篇paper有人搜得到么。？或者源作者说的博客，怎么都进不去

请问预训练模型的源字体和目标字体分别是什么呢？

不管split_ratio为多少，train.obj都是一个空文件。。。

您好，非常感谢您分享这个伟大项目，我在执行train.py时，遇到下面bug了，可以告知怎么解决吗？
unpickled total 13553 examples
Traceback (most recent call last):
File "train.py", line 126, in
main()
File "train.py", line 100, in main
for bid, batch in enumerate(dataloader):
File "/usr/python3.8/lib/python3.8/site-packages/torch/utils/data/dataloader.py", line 363, in next
data = self._next_data()
File "/usr/python3.8/lib/python3.8/site-packages/torch/utils/data/dataloader.py", line 403, in _next_data
data = self._dataset_fetcher.fetch(index) # may raise StopIteration
File "/usr/python3.8/lib/python3.8/site-packages/torch/utils/data/utils/fetch.py", line 44, in fetch
data = [self.dataset[idx] for idx in possibly_batched_index]
File "/usr/python3.8/lib/python3.8/site-packages/torch/utils/data/utils/fetch.py", line 44, in
data = [self.dataset[idx] for idx in possibly_batched_index]
File "/source/code/panxy/src/FontGeneration/zi2zi/data/dataset.py", line 35, in getitem
img_A, img_B = self.process(item[1])
File "/source/code/panxy/src/FontGeneration/zi2zi/data/dataset.py", line 102, in process
img_A = self.transform(img_A)
File "/usr/python3.8/lib/python3.8/site-packages/torchvision/transforms/transforms.py", line 212, in call
return F.normalize(tensor, self.mean, self.std, self.inplace)
File "/usr/python3.8/lib/python3.8/site-packages/torchvision/transforms/functional.py", line 298, in normalize
tensor.sub(mean).div(std)
RuntimeError: output with shape [1, 256, 256] doesn't match the broadcast shape [3, 256, 256]

train.py阶段

Traceback (most recent call last):
File "train.py", line 128, in
main()
File "train.py", line 79, in main
model.setup()
File "F:\zi2zi-pytorch-master\model\model.py", line 59, in setup
init_net(self.netG, gpu_ids=self.gpu_ids)
File "F:\zi2zi-pytorch-master\utils\init_net.py", line 15, in init_net
assert(torch.cuda.is_available())
AssertionError

Traceback (most recent call last):
File "package.py", line 76, in
with open(dst_json, 'r', encoding='utf-8') as fp:
TypeError: expected str, bytes or os.PathLike object, not NoneType

最少多少个自己手写字符就可以用来训练模型生成其它字符的手写效果？

我运行你的开源代码，感觉字体生成的效果不如zi2zi的了。效果不是很好。是我哪里的设置不对么。

我后来又在你的代码基础上增加了vgg loss ，生成的效果也不是很好。