CMX (RGB+NP++)

Reported on 5 benchmarks across 1 task · 1 paper · 2 SOTA

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Computer Vision5 results

Image Manipulation LocalizationonColumbia
Average Pixel F1(Fixed threshold)· 2022-03-09
0.884
best: 0.888 (Early Fusion)
SOTA
CMX: Cross-Modal Fusion for RGB-X Semantic Segmentation with Transformers arXiv:2203.04838
Image Manipulation LocalizationonDSO-1
Average Pixel F1(Fixed threshold)· 2022-03-09
0.895
best: 0.93 (TruFor)
SOTA
CMX: Cross-Modal Fusion for RGB-X Semantic Segmentation with Transformers arXiv:2203.04838
Image Manipulation LocalizationonCOVERAGE
Average Pixel F1(Fixed threshold)· 2022-03-09
0.577
best: 0.663 (Early Fusion)
CMX: Cross-Modal Fusion for RGB-X Semantic Segmentation with Transformers arXiv:2203.04838
Image Manipulation LocalizationonCasia V1+
Average Pixel F1(Fixed threshold)· 2022-03-09
0.761
best: 0.791 (CMX (RGB+SRM))
CMX: Cross-Modal Fusion for RGB-X Semantic Segmentation with Transformers arXiv:2203.04838
Image Manipulation LocalizationonCocoGlide
Average Pixel F1(Fixed threshold)· 2022-03-09
0.516
best: 0.585 (CMX (RGB+SRM))
CMX: Cross-Modal Fusion for RGB-X Semantic Segmentation with Transformers arXiv:2203.04838