Here are several mistakes on the correspondance in "our485", like the 79.png, 735.png. The order of low and high images in "our485" do not match, like the 79.png and 111.png. But I correct it ,the decom net train result seem to bee worse! How do you solve it ? #6

changruowang · 2019-12-12T02:36:32Z

No description provided.

zhangyhuaee · 2019-12-12T05:45:43Z

The original LOLdataset has some mistakes. We rearrange the dataset and add several all-zero images to get better results. You can train it again with our new dataset. Thank you!

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Here are several mistakes on the correspondance in "our485", like the 79.png, 735.png. The order of low and high images in "our485" do not match, like the 79.png and 111.png. But I correct it ,the decom net train result seem to bee worse! How do you solve it ? #6

Here are several mistakes on the correspondance in "our485", like the 79.png, 735.png. The order of low and high images in "our485" do not match, like the 79.png and 111.png. But I correct it ,the decom net train result seem to bee worse! How do you solve it ? #6

changruowang commented Dec 12, 2019

zhangyhuaee commented Dec 12, 2019 •

edited

Loading

Here are several mistakes on the correspondance in "our485", like the 79.png, 735.png. The order of low and high images in "our485" do not match, like the 79.png and 111.png. But I correct it ,the decom net train result seem to bee worse! How do you solve it ? #6

Here are several mistakes on the correspondance in "our485", like the 79.png, 735.png. The order of low and high images in "our485" do not match, like the 79.png and 111.png. But I correct it ,the decom net train result seem to bee worse! How do you solve it ? #6

Comments

changruowang commented Dec 12, 2019

zhangyhuaee commented Dec 12, 2019 • edited Loading

zhangyhuaee commented Dec 12, 2019 •

edited

Loading