北邮-刘畅

How does it work with "Generator3DLUT_identity"

2

### In the model.py and the code for training,  ### What is the difference of the "Generator3DLUT_identity" and "Generator3DLUT_zero" ? ### At first, in my opinion, the parameters of...

The result is not satisfactory because PSNR reaches about 23? What Happened?

3

I have managed to run the code and I test on the same test datasets, only to find disappointing result. Is there anyone encoutering the same problem. Listed is my...

Any plan to release the pretrained weight file?

1

As stated in the title.

请问动态 shape 下的空间如何申请呢？

1

请问，大佬知道 dynamic shape 下的 getWorkspaceSize 如何写吗？静态的很好判断，但目前个人自定义的算子卡在了这里，还望大佬指点！

你好！我下载源码之后，直接把 pair 数据放在该放的位置，直接训练，训练好几个小时了，psnr 涨的好慢？是我哪里需要调什么吗？ 

使用官方的 5 层 CNN，训练结果 使用 ResNet18，训练结果 使用 EfficientNet_b0， 使用 EfficientNet_v2 使用 neighborhood attention network， 请问有哪位遇到过类似的问题吗？

Although the enhanced results is somehow underexposed, it's good enough. However, more work may need to be done upon speed.

There is a mistake while nms!

1

I have trained on PASCAL-VOC2007 with the pretrained resnet50. However, while test a single image, problem occured！  **Multiply results are the same and this is no coincidence.**  ...

你好！ 十分感谢你的无私开源，让我等小白好入门。 我按照步骤成功训练后，detect 达到的效果有点差？不知是否有什么 trick ? 如果可以公开 pretrained 模型就更好了！