Semantic Segmentation on Cityscapes (val) (Standard mIoU)
79mIoUPEM
Evaluation results
| Method | Links | |
|---|---|---|
| PEMBackbone=STDC2, Resolution=1024 × 2048, FPS=14.22026.01 | 79 | |
| RDRNetBackbone=-, Resolution=1024 × 2048, FPS=24.22026.01 | 78.9 | |
| GAINBackbone=ResNet-18, Resolution=1024 × 2048, FPS=22.32026.01 | 78.8 | |
| SFNetBackbone=ResNet-18, Resolution=1024 × 2048, FPS=182026.01 | 78.3 | |
| PEMBackbone=STDC1, Resolution=1024 × 2048, FPS=16.62026.01 | 78.3 | |
| GAINBackbone=DF-2, Resolution=1024 × 2048, FPS=43.82026.01 | 78.3 | |
| HyperSeg-SBackbone=EfficientNet-B1, Resolution=768 × 1536, FPS=16.12026.01 | 78.2 | |
| BiDGANet-LBackbone=-, Resolution=1024 × 2048, FPS=23.52026.01 | 77.9 | |
| STDC2Backbone=STDC2, Resolution=768 × 1536, FPS=97, TensorRT acceleration=true2026.01 | 77 | |
| HyperSeg-MBackbone=EfficientNet-B1, Resolution=512 × 1024, FPS=36.92026.01 | 76.2 | |
| BiSeNetV2Backbone=-, Resolution=512 × 1024, FPS=47.3, TensorRT acceleration=true2026.01 | 75.8 | |
| SwiftNetBackbone=ResNet-18, Resolution=1024 × 2048, FPS=39.92026.01 | 75.4 | |
| BiDGANet-BBackbone=-, Resolution=1024 × 2048, FPS=39.82026.01 | 75.2 | |
| BiSeNetV1Backbone=ResNet-18, Resolution=768 × 1536, FPS=65.52026.01 | 74.8 | |
| STDC1Backbone=STDC1, Resolution=768 × 1536, FPS=126.7, TensorRT acceleration=true2026.01 | 74.5 | |
| BiSeNetV1Backbone=Xception-39, Resolution=768 × 1536, FPS=105.82026.01 | 69 | |
| Fast SCNNBackbone=-, Resolution=1024 × 2048, FPS=123.52026.01 | 68.6 | |
| ESPNetV2Backbone=-, Resolution=512 × 1024, FPS=61.92026.01 | 66.4 |