alexliyang

when I use “BASICSR_EXT=True pip install basicsr“ to install basicsr， then I use the recommended train command， but always give error msg like :" NameError: name 'fused_act_ext' is not defined...

这个是直接用1024*1024的图像训练了吧，resieze到512*512可以开始训练，但是是否应该resize呢？

知道这个错误的原因了，当输入图片的大小是 3000*4000时， [1,3,1,1,16] 张量和[1,3,3000,4000,16]的张量计算乘法时，内存出现了错误。 期望能把内存这块的在优化一下

I have not got the annotation files. which dataset do you use? and do you have the tool to label samples to generate the annotation file?

哦， 多谢了 在您的论文之后，有两篇类似的论文： ---Real-time Image Enhancer via Learnable Spatial-aware 3D Lookup Tables ---Learning Pixel-Adaptive Weights for Portrait Photo Retouching 似乎是要解决在不同空间位置的相同颜色可以映射为不同颜色的问题，这样对于不同大小的图片经过了resize后效果不知道怎么样？ 另外，如何评价使用自适应lut表处理后的图像质量呢？ 用PSNR/SSIM？ 怎么主观感觉处理后的图像的清晰度（锐度）相比原张有所下降？ 多谢

@tirtile ：when is it the loss value you stop the training?

thx a lot Now I want to train a classical model, samples do not use bbox , only classic labels, how to use your loss function? Thanks

OK, thanks, I will use this loss to train my model. and another question: my class_num is 10, whether I should add 1 to class_num i.e. class_num = 11 when...

@wubaoyuan thanks, although I do not understand your loss function , now I copy it and use it to train my model, I find that ,the every step loss value...

：） another question: how to keep the unshadowed area color unchanged after shadow removing, when I remove the shadow, other area colors will be changed.