Single Image Reflection Removal on Five Real-world Datasets Average (test)
27.21PSNRDPIT
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DPITVenue=Not applicable, Trainable Params(M)=131.54, FLOPs(G)=191.35, Pre-trained=false2025.05 | 27.21 | 0.924 | |
| RDNetVenue=CVPR 2025, Trainable Params(M)=315.89, FLOPs(G)=183.9, Pre-trained=true2025.05 | 26.72 | 0.917 | |
| DSITVenue=NeurIPS 2024, Trainable Params(M)=131.76, FLOPs(G)=233.09, Pre-trained=true2025.05 | 26.71 | 0.918 | |
| LLCNVenue=Not applicable, Trainable Params(M)=99.44, FLOPs(G)=24.1, Pre-trained=false2025.05 | 26.12 | 0.899 | |
| DSRNetVenue=ICCV 2023, Trainable Params(M)=123.67, FLOPs(G)=276.59, Pre-trained=true2025.05 | 25.84 | 0.91 | |
| Zhu et al.Venue=CVPR 2024, Trainable Params(M)=19.67, FLOPs(G)=12.33, Pre-trained=true2025.05 | 25.6 | 0.899 | |
| HGNetVenue=TNNLS 2023, Trainable Params(M)=14.5, FLOPs(G)=303.75, Pre-trained=true2025.05 | 24.78 | 0.895 | |
| Dong et al.Venue=ICCV 2021, Trainable Params(M)=10.93, FLOPs(G)=256.11, Pre-trained=true2025.05 | 24.35 | 0.896 | |
| Li et al.Venue=CVPR 2020, Trainable Params(M)=21.61, FLOPs(G)=300.35, Pre-trained=true2025.05 | 24.11 | 0.881 |