List your company
ResearchFormerly WizWand

Semantic Segmentation on GTA5

60.7mIoUCMFormer
25.142434.373743.60552.8363Jul 1, 2023Sep 19, 2023Dec 9, 2023Feb 28, 2024May 19, 2024Aug 8, 2024Oct 28, 2024
Updated 5mo ago

Evaluation results

MethodLinks
CMFormerBackbone=Swin-L, Trained on Cityscapes=true2023.0760.7
HGFormerBackbone=Swin-L, Trained on Cityscapes=true2023.0759.4
CMFormerBackbone=Swin-B, Trained on Cityscapes=true2023.0758.11
Mask2FormerBackbone=Swin-L, Trained on Cityscapes=true2023.0757.8
Mask2FormerBackbone=Swin-B, Trained on Cityscapes=true2023.0755.05
HGFormerBackbone=Swin-T, Trained on Cityscapes=true2023.0751.3
Mask2FormerBackbone=Swin-T, Trained on Cityscapes=true2023.0750.6
SHADEBackbone=ResNet50, Trained on Cityscapes=true2023.0748.61
SAWBackbone=ResNet50, Trained on Cityscapes=true2023.0747.28
WildNetBackbone=ResNet50, Trained on Cityscapes=true2023.0747.01
DIRLBackbone=ResNet50, Trained on Cityscapes=true2023.0746.52
GTRBackbone=ResNet50, Trained on Cityscapes=true2023.0745.79
IternormBackbone=ResNet50, Trained on Cityscapes=true2023.0745.73
DRPCBackbone=ResNet50, Trained on Cityscapes=true2023.0745.62
IBNBackbone=ResNet50, Trained on Cityscapes=true2023.0745.06
ISWBackbone=ResNet50, Trained on Cityscapes=true2023.0745
IWBackbone=ResNet50, Trained on Cityscapes=true2023.0744.87
ULDASource Domain=Cityscapes, Prompt=driving in a game, Backbone=CLIP-ResNet-50, Base Model=DeepLabv3+2024.0442.91
PØDAAdaptation setting=0-shot, Prompt=driving in a game2024.1041.07
PØDASource domain=Cityscapes (CS), Target prompt (P)=driving in a game, Zero-shot domain adaptation=true2024.1041.07
PIDAAdaptation setting=1-shot, Target image usage=single target image2024.1041.02
PØDASource Domain=Cityscapes, Prompt=driving in a game, Backbone=CLIP-ResNet-50, Base Model=DeepLabv3+2024.0440.77
CMFormerTrained on=SYNTHIA (S)2023.0740.65
AdaINAdaptation setting=1-shot, Target image usage=single target image2024.1040.56
source-onlySource Domain=Cityscapes, Prompt=driving in a game, Backbone=CLIP-ResNet-50, Base Model=DeepLabv3+2024.0439.59
source-only2024.1039.59
source-onlySource domain=Cityscapes (CS), Target prompt (P)=driving in a game, Zero-shot domain adaptation=true2024.1039.59
CLIPstylerSource Domain=Cityscapes, Prompt=driving in a game, Backbone=CLIP-ResNet-50, Base Model=DeepLabv3+2024.0438.73
CLIPstylerSource domain=Cityscapes (CS), Target prompt (P)=driving in a game, Zero-shot domain adaptation=true2024.1038.73
SAWTrained on=SYNTHIA (S)2023.0729.16
DRPCTrained on=SYNTHIA (S)2023.0728.75
GTRTrained on=SYNTHIA (S)2023.0728.02
ISWTrained on=SYNTHIA (S)2023.0727.68
IBNTrained on=SYNTHIA (S)2023.0726.9
IWTrained on=SYNTHIA (S)2023.0726.51