csjliang / ppr10k Goto Github PK

after run sh setup.sh successfully, I run 'import trilinear' and then arise the following error:

import trilinear
Traceback (most recent call last):
File "", line 1, in
ImportError: /usr/local/python3.6.5/lib/python3.6/site-packages/trilinear-0.0.0-py3.6-linux-x86_64.egg/trilinear.cpython-36m-x86_64-linux-gnu.so: undefined symbol: _ZN6caffe26detail36_typeMetaDataInstance_preallocated_4E

my environment as below:
ubuntu 16.04
python 3.6
pytorch 1.1
cuda 10.0

Hello. Your research's result is very nice! I have one question.
The models_x.py imports trilinear, but I only found the trilinear_cpp folder.
How do I import trilinear.py?

作者你好，十分感谢这个出色的数据集！我想问一问你们有png格式的可以下载吗，raw格式确实是有点大，下载很慢&存储不方便

Thank you for your great work.

I want to download your dataset, but download link (Google drive, OneDrive) is not available.

Where I can get them?

Thank you.

请问released模型分别对应论文的哪些结果。
两个released模型的PSNR似乎都远远超过了Table 1给出的PSNR

I download the file and unzip them, then execute instructions with the data path end with "PPR0K_all_files_11161_zip", but the dataloder is always empty during debugging, I tried other paths but still couldn't read the data, could you please give me a complete example, including the instruction to read data?

in train_val_images_tif_360p, each scene has six different version photo under source_aug_6, what's the difference between them and which one should be used for training model?

请问能否release 3DLUT without HRP without GRC的三个模型 (abc)，貌似pretrained-model下载链接里没有提供

Are there any plans to release the portrait segmentation algorithm as well?

你好： 我想自己做一些数据集，能否详细描述下如何做出您那样的数据集呢？

the validation.py proposed resnet18_224 as a pretrained test model, how about the classify acc of Classifier2 in model_x.py ?

Thanks for your excellent work on this paper. I have read your tips in dataset_usage.md but still have some questions about the processing of HR images. The file does not mention how I should use the HR images for the testing part of the experiment and thus get the PSNR, Eab, etc., data for the HR images.
Do I need to use Adobe Lightroom to convert the original image format composed of RAW and XMP to TIF and retain the original image size?
I would really appreciate your answer.

你好：
在进行训练得到一个较好的模型后做测试，发现有些情况下的color banding的现象比较严重，这种情况如何解决？

上传不了图片，现象就是对于相对纯色的情况下，比如纯白、纯黄等，会出现条带装的色带。

谢谢

Hi,
Thanks for sharing your codes and datasets.
However, during the process of preparing HR datasets, I find out mismatch between raw files and xmp files.
Though xmp files of target A are complete, xmp files of target B and C are mismatched with raw data.
Would you mind checking xmp files again and update them? Your help would be appreciated.

Regards,
Huimin

Hi, I have tried several days to download raw.zip but it always shows the download quota is exceeded. Could you share another Google drive link?

如何保证lut表中的值在[0,1]之间？ 而且还存在 组合后的lut表的值是否在[0,1]之间，当不在这个区间时，特别是负值时，应该如何处理？

good job , the 360p is small for my trainning net, how to get the srgb fullsize source and target ? thanks

Hello,

This database is really interesting.

I tried to download the database but on google drive, it gives me a quota exceeded error, and on onedrive the download button doesn't respond. Any solutions?

Thank you very much,

Best

Could you please tell me the training cost for this project, i.e. how many GPUs, how long to finish the training process?