LLFormer
LLFormer copied to clipboard
TaoWangzj
The code release of paper "AAAI Ultra-High-Definition Low-Light Image Enhancement: A Benchmark and Transformer-Based Method", AAAI 2023
感谢您对这个优秀的工作的分享，对于FiveK数据集有一个疑问 将专家C的文件作为GT，那么对应的低光照图像是怎么得到的呢？与DeepUPE论文中的转化流程一样吗？ 因为，我发现您公布的数据集中的低光照图像与RetinexFormer中提供的数据集中的图像不一致。
May I ask the spec of the server? Does this project require a high-end server?
Hello thank you for your effort. When I load the model (Adobe FiveK/model_bestPSNR.pth), it prompt this error. How can I resolve them? ( I changed to torch.load(weights, map_location="cpu")) Thank you
Error when downloading the model, it may be that the model does not match the code, can you provide the correct network architecture
Hello, in the code you released I found that the qkv matrix is generated using a 1*1 convolution and 2 depth-conv, but in the structure picture there is only a...
Hello! I tried to use the configuration and code you released and repeated the training for many times. However, the results on LOLv1 cannot be reproduced. The highest PSNR is...
