Comments (21)
We are currently checking the license issue to make the dataset public.
Please wait a little.
from ugatit.
The selfie dataset appears to be this one: https://www.crcv.ucf.edu/data/Selfie/
Since the paper says they scraped anime-planet themselves to get the anime images, it seems like we'll have to yield to them for their dataset. Alternatively, there appears to be a dataset made available here that might work. http://www.nurs.or.jp/~nagadomi/animeface-character-dataset/
from ugatit.
-
We released 50 epoch and 100 epoch checkpoints so that people could test more widely.
-
Also, We published the selfie2anime datasets we used in the paper.
-
And, we fixed code in smoothing
-
In the test image, I recommend that your face be in the center.
from ugatit.
数据集和模型下载
==========================
链接:https://pan.baidu.com/s/1dP1mXuU-rA9dPvFe8YS8jQ
提取码:k6rc
已添加checkpoint文件
from ugatit.
You can also try this anime_faces dataset that I parsed from the tfrecords of twingan project.
As for a replacement of selfie dataset, you can try celebA and even CelebAMask-HQ(this can help you generate a high-res portraits dataset without backgrounds)
from ugatit.
Note that the nagadomi faces are also on Kaggle, and there's at least 2 anime face datasets derived from Danbooru2018: "Danbooru 2018 Anime Character Recognition Dataset" and SeePrettyFace.com: face dataset, in addition to the ones I made for my StyleGAN/BigGAN projects but haven't formally released.
from ugatit.
Here's my dataset:
And I did some preparations:
According to the paper 5.2. Dataset, I wrote this script and selected 3400 female selfies(but the selfie labels seems contains some errors so there's still some male selfies) as trainset and 100 as testset. relatively, I choose the biggest 3500 anime pictures as anime trainset and testset.
You can download this 2*3500 dataset from here.
This dataset still can be improved, especially the selfie dataset. If the angle and position can be more corresponding with the anime, I think the preference will be better. as a matter of fact, these samples manually picked above are those selfies relatively good.
from ugatit.
Here's my dataset:
And I did some preparations:
According to the paper 5.2. Dataset, I wrote this script and selected 3400 female selfies(but the selfie labels seems contains some errors so there's still some male selfies) as trainset and 100 as testset. relatively, I choose the biggest 3500 anime pictures as anime trainset and testset.
You can download this 2*3500 dataset from here.
This dataset still can be improved, especially the selfie dataset. If the angle and position can be more corresponding with the anime, I think the preference will be better. as a matter of fact, these samples manually picked above are those selfies relatively good.
That's very nice summary, I think the dataset the author use was not so good for selfies, it's so disperse. As you said, maybe we should train with specific selfies, such as registration photo?
So, did you retrain the model with your new datasets? How's the result look like?
from ugatit.
Dataset for Anime images 5K from http://www.seeprettyface.com
from ugatit.
https://www.gwern.net/Danbooru2018 with some filters would be a great anime face dataset.
from ugatit.
Note that the nagadomi faces are also on Kaggle, and there's at least 2 anime face datasets derived from Danbooru2018: "Danbooru 2018 Anime Character Recognition Dataset" and SeePrettyFace.com: face dataset, in addition to the ones I made for my StyleGAN/BigGAN projects but haven't formally released.
The one of SeePrettyFace.com are a pain in the ass to download. Require's me to download baidu's downloader and register in Chinese (I believe).
from ugatit.
Hi, I have a question, I run python main.py --dataset selfie2anime but I not sure if it is working.
I know it is weird quiestion, I new in these
from ugatit.
Hi, I have a question, I run python main.py --dataset selfie2anime but I not sure if it is working.
I know it is weird quiestion, I new in these
'cuz the dataset is not yet included here
from ugatit.
You can also try this anime_faces dataset that I parsed from the tfrecords of twingan project.
As for a replacement of selfie dataset, you can try celebA and even CelebAMask-HQ(this can help you generate a high-res portraits dataset without backgrounds)
In the directory structure of README.md doesn't it say that the training set should be jpg or png's? (haven't read the source code).
from ugatit.
- How to setup a checkpoint? it keeps saying loading checkpoint failed.
- i'm using GTX1060 6GB. Should I readjust batch_size to avoid 'resource exhausted' issue?
Thx
from ugatit.
@sdy0803
You can change checkpoint save frequency in main.py. If you have any checkpoint saved, it will probably load successfully.
from ugatit.
@sdy0803
You can change checkpoint save frequency in main.py. If you have any checkpoint saved, it will probably load successfully.
The thing is that i find my GTX 1060 6GB cannot even run through the whole training progress with a miserable training dataset size of 140 pictures
from ugatit.
According to #3
Their setup was
One NVIDIA V100 32GB GPU.
It takes 4 days.
So it is not so surprising that it will take much longer with 1060.
I finished the training by significantly reducing number of epoch and training iteration (obviously produced terrible results). If you just want to finish training, then you can try the same.
from ugatit.
thx. let me give it one more try
from ugatit.
You can also try this anime_faces dataset that I parsed from the tfrecords of twingan project.
As for a replacement of selfie dataset, you can try celebA and even CelebAMask-HQ(this can help you generate a high-res portraits dataset without backgrounds)
Thanks for your sharing. Did you retrain the model with your anime dataset? Since they are in low resolution, did it affects?
from ugatit.
Dataset for Anime images 5K from http://www.seeprettyface.com
link of google drives is dead, can you update the link?
from ugatit.
Related Issues (20)
- what's the difference between local and global attention maps HOT 2
- Why HOT 2
- CAM loss is degrading the results HOT 6
- saturated result image HOT 1
- resume train HOT 1
- Umm what am I looking at?!?!? HOT 5
- I publish a PyTorch version, welcom to follow.
- Failed to find a checkpoint
- "Direction" of the model
- About CAM loss
- 官方模型解压出错,文件损坏。有小伙伴遇到同样问题吗? HOT 2
- Pre-trained model provided on googl drive looks broken HOT 1
- Augmentation probability
- Testing does not work!
- In paper, "5.3.2 AdaLIN analysis", how to use AdaIN for comparative experiments? HOT 1
- 这个做法有什么输出结果吗?
- Is there checkpoints after more epoches?
- gril
- How much gpu memory is required to run the full UGATIT code?
- I'm having a lot of issues when trying to run with Tensorflow2x
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 ugatit.