Understanding Dark Scenes by Contrasting Multi-Modal Observations

Benchmark Model Rank Results
semantic-segmentation-on-llrgbd-syntheticSMMCL (SegNeXt-B)#1mIoU: 68.76
semantic-segmentation-on-llrgbd-syntheticSMMCL (SegFormer-B2)#2mIoU: 67.77
semantic-segmentation-on-llrgbd-syntheticSMMCL (ResNet-101)#5mIoU: 64.40
semantic-segmentation-on-nyu-depth-v2SMMCL (SegNeXt-B)#21Mean IoU: 55.8%
semantic-segmentation-on-nyu-depth-v2SMMCL (SegFormer-B2)#29Mean IoU: 53.7%
semantic-segmentation-on-nyu-depth-v2SMMCL (ResNet-101)#37Mean IoU: 52.5%