Semantic Segmentation on COCO-Stuff 10K
54.56mIoUPlainSeg-Hier
Evaluation Results
| Method | Links | |
|---|---|---|
| PlainSeg-Hierbackbone=BEiTv2-L, PRM=326M, test time=188ms2023.10 | 54.56 | |
| SegViTbackbone=BEiTv2-L, PRM=344M, test time=158ms2023.10 | 52 | |
| SegViTBackbone=ViT-Large, GFLOPs=373, FPS=44.52025.09 | 46.7 | |
| SegViT + DToPBackbone=ViT-Large, GFLOPs=290, FPS=152025.09 | 46.6 | |
| SegViT + CTSBackbone=ViT-Large, GFLOPs=251, FPS=402025.09 | 46.2 | |
| SegViT + STEP@[8,16]τ-6899Backbone=ViT-Large, GFLOPs=173, FPS=182025.09 | 46 | |
| SegViT + STEP@[18]τ-6899Backbone=ViT-Large, GFLOPs=201, FPS=302025.09 | 46 | |
| SegViT + CTS & DToPBackbone=ViT-Large, GFLOPs=210, FPS=172025.09 | 45.4 | |
| SegViT + STEP@[8,16]τ-4999Backbone=ViT-Large, GFLOPs=150, FPS=202025.09 | 45.3 | |
| SegViT + STEP@[18]τ-4999Backbone=ViT-Large, GFLOPs=177, FPS=292025.09 | 45.1 | |
| TeacherBackbone=ResNet182022.05 | 33.1 | |
| TeacherBackbone=MobilenetV22022.05 | 33.1 | |
| Target-aware TransformerBackbone=ResNet182022.05 | 28.75 | |
| TaTBackbone=ResNet182022.05 | 28.75 | |
| Target-aware TransformerBackbone=MobileNetV22022.05 | 28.05 | |
| TaTBackbone=MobilenetV22022.05 | 28.05 | |
| OverhaulBackbone=ResNet182022.05 | 27.86 | |
| ICKDBackbone=ResNet182022.05 | 27.22 | |
| OverhaulBackbone=MobileNetV22022.05 | 26.96 | |
| KRBackbone=ResNet182022.05 | 26.73 | |
| ICKDBackbone=MobileNetV22022.05 | 26.64 | |
| KRBackbone=MobilenetV22022.05 | 26.63 | |
| StudentBackbone=ResNet182022.05 | 26.33 | |
| StudentBackbone=MobilenetV22022.05 | 26.29 |