liuqk3 / PUT

Paper 'Transformer based Pluralistic Image Completion with Reduced Information Loss' in TPAMI 2024 and 'Reduce Information Loss in Transformers for Pluralistic Image Inpainting' in CVPR2022
MIT License
173 stars 15 forks source link

inference error #13

Closed chencc5956 closed 1 year ago

chencc5956 commented 1 year ago

您好,非常感谢您的论文,给了我很大启发。我在复现代码的过程中发生了一些问题。 ①我是想做一个简单的测试,所以只用了一张celeba-hq里的图片和一张mask图片进行测试,不知道是不是因为数据的存放路径有问题,一直是显示这个error。请问可以帮忙解惑吗?

image

②另外,一直没有找到scripts文件夹里/tools/dataset_tool.py这个文件。

谢谢您!!!

liuqk3 commented 1 year ago

Thanks for your interests!

You should change the configurations for you validation data. The name and image_list_file are the key to find the image correctly.

https://github.com/liuqk3/PUT/blob/7de8ce0ada1e63e8c5300857a463d68380e142f0/configs/put_cvpr2022/ffhq/transformer_ffhq.yaml#L115 https://github.com/liuqk3/PUT/blob/7de8ce0ada1e63e8c5300857a463d68380e142f0/configs/put_cvpr2022/ffhq/transformer_ffhq.yaml#L116

chencc5956 commented 1 year ago

OK, I solved this problem. Thanks for your help.