XunpengYi / Diff-Retinex

Official Code of the Diff-Retinex: Rethinking Low-light Image Enhancement with A Generative Diffusion Model (ICCV2023)
MIT License
20 stars 0 forks source link

GT-means 测试问题 #3

Open RenzhiHu111 opened 3 weeks ago

RenzhiHu111 commented 3 weeks ago

您好,我想问一下有哪几篇低光增强算法是用了下面GT-means的测试方法

some method get very high PSNR (more than 27), check this in the code

    if args.use_gtmeans is True:
        normal_I = val_data['high'].detach().cpu().numpy().squeeze(0)*255.0
        normal_I_means = np.mean(normal_I)
        pred_I_means = np.mean(I)
        I = I * normal_I_means / pred_I_means
XunpengYi commented 3 weeks ago

您可以参照对应算法的实现代码。例如,KinD和LLFlow都实现了相应的接口。