shaosifan / transenet Goto Github PK

When to open source code!
thks a lot.

Waiting for your kindly reply,
Regards,

你好，想问一下您有AID数据集下采样后的百度网盘吗。AID数据集是jpg文件，我下采样后保存为jpg文件就自动压缩了，所以下采样后是保存为png文件吗。

您好！读了您的论文之后我非常想尝试一下TransENet模型来适用于我现阶段的实验。
目前我已经收集了LR/HR影像对，但这些遥感影像尺寸较大为tif文件并且有24个波段，请问应该如何制作patches并且适用于您这里的网络呢？
期待您的回复！非常感谢！

Hi,Could you support the pretrained models about TransENet?Thanks.
I try to retrain the model,but there are some questions. The test results are so bad.

在generate_mod_LR_bic文件中用到imresize_np并没有提供。可以麻烦补充下吗，谢谢

请问您测试结果是在Y通道测试还是在RGB通道测试

你好，我想问一下，论文里的数据集是按照您百度网盘提供的数据集进行训练的吗，您百度网盘提供的ucmerd数据集，验证集有105张，是按照10%的训练集划分的，不过今天才看到论文里面写的是20%的训练集作为验证集。所以想问一下。

Thank you for your excellent work, I would like to ask if you can provide me with a description of the number of model parameters. When I tested the x4 factor with an image of size (1,3,48,48), I found that it has 37.5 M. I would like to know if this is the correct number of parameters.
Looking forward to your reply!!!

你好，文章中的视觉效果对比图是需要retrain这些模型的吗 ？能否提供这些方法在AID和UCMerced数据集上复原的图像？ 谢谢

您好，AID数据集的rgb_mean和rgb_std ，您有计算吗？我在代码里只看到rgb_mean = (0.4916, 0.4991, 0.4565) # UCMerced data
rgb_std = (1.0, 1.0, 1.0)

我想问一下可以提供一下这个裁剪的大小，还有坐标在哪里，然后就是通过什么方法裁剪的，非常感谢作者。

Excuse me, I can't find loss.pt after running demo_train, where did this loss weight come from?

Thanks for your great job introducing Transformer into remote sensing image SR!
About the dataset used in the paper, could you provide it directly? Is the UCMereced dataset split into two parts randomly?