Zero-DCE
Zero-DCE copied to clipboard
Li-Chongyi
About the result
Hello, I have obtained 786 pairs of images in the part2 data set of SICE you mentioned. The PSNR obtained by comparing the results obtained with the trained model you provided is always 0.1 worse than that in the paper. What is wrong?
Hello, I have obtained 786 pairs of images in the part2 data set of SICE you mentioned. The PSNR obtained by comparing the results obtained with the trained model you provided is always 0.1 worse than that in the paper. What is wrong?
I think it’s because of the problem of data set selection. The paper said that if there are 7 pictures, choose 3 pictures, and 9 pictures choose 4 pictures, but the paper did not clearly state the situation of 5, 6, 8 pictures.
你好，你说的SICE的part2数据集中，我已经得到了786对图片。与你提供的训练模型对比得到的结果，得到的 PSNR 总是比论文中的差 0.1。怎么了？
我认为是因为数据集选择的问题。论文说如果有7张图片，选择3张图片，9张图片选择4张图片，但是论文没有明确说明5、6、8张图片的情况。
您好！方便提供下PSNR和SSIM的代码吗？谢谢
您好，我在您提到的 SICE 的第 786 部分数据集中获得了 2 对图像。通过将获得的结果与您提供的训练模型进行比较而获得的 PSNR 总是比论文中的差 0.1。怎么了？
我认为这是因为数据集选择的问题。该报称，如果有7张图片，选择3张图片，9张图片选择4张图片，但该报没有明确说明5张、6张、8张图片的情况。
Hello, how do you choose between 5, 6, 8? I chose 2, 2, 3 of them, but the PSNR calculated was 0.34 less than the author's
