Synthetic To Real Translation On Synthia To 1

Métriques

MIoU (13 classes)

MIoU (16 classes)

Résultats

Résultats de performance de divers modèles sur ce benchmark

Nom du modèle	MIoU (13 classes)	MIoU (16 classes)	Paper Title	Repository
MetaCorrection(ResNet-101)	52.5	45.1	MetaCorrection: Domain-aware Meta Loss Correction for Unsupervised Domain Adaptation in Semantic Segmentation
TransDA-B	66.3	59.3	Smoothing Matters: Momentum Transformer for Domain Adaptive Semantic Segmentation
DACS(ResNet-101)	54.81	48.34	DACS: Domain Adaptation via Cross-domain Mixed Sampling
DCF	75.9	69.3	Transferring to Real-World Layouts: A Depth-aware Framework for Scene Adaptation
HRDA	72.4	65.8	HRDA: Context-Aware High-Resolution Domain-Adaptive Semantic Segmentation
EHTDI	64.6	57.8	Exploring High-quality Target Domain Information for Unsupervised Domain Adaptive Semantic Segmentation
AdaptSegNet(Multi-level)	46.7	-	Learning to Adapt Structured Output Space for Semantic Segmentation
SePiCo	71.4	64.3	SePiCo: Semantic-Guided Pixel Contrast for Domain Adaptive Semantic Segmentation	-
CLUDA+HRDA	-	67.2	CLUDA : Contrastive Learning in Unsupervised Domain Adaptation for Semantic Segmentation
FADA(ResNet-101)	52.5	45.2	Classes Matter: A Fine-grained Adversarial Approach to Cross-domain Semantic Segmentation
MIC	74.0	67.3	MIC: Masked Image Consistency for Context-Enhanced Domain Adaptation
CAMix (w DAFormer)	69.2	-	Context-Aware Mixup for Domain Adaptive Semantic Segmentation
CAG-UDA	52.6	44.5	Category Anchor-Guided Unsupervised Domain Adaptation for Semantic Segmentation
ProDA(ResNet-101)	62.0	55.5	Prototypical Pseudo Label Denoising and Target Structure Learning for Domain Adaptive Semantic Segmentation
CAMix (ResNet 101)	59.7	-	Context-Aware Mixup for Domain Adaptive Semantic Segmentation
CLAN	47.8	-	Taking A Closer Look at Domain Shift: Category-level Adversaries for Semantics Consistent Domain Adaptation
Uncertainty + Adaboost (ResNet-101)	57.5	50.4	Adaptive Boosting for Domain Adaptation: Towards Robust Predictions in Scene Segmentation
G2L	64.4	56.8	G2L: A Global to Local Alignment Method for Unsupervised Domain Adaptive Semantic Segmentation	-
IAST(ResNet-101)	57.0	49.8	Instance Adaptive Self-Training for Unsupervised Domain Adaptation
ITEN	-	46.45	Exploiting Image Translations via Ensemble Self-Supervised Learning for Unsupervised Domain Adaptation	-

0 of 36 row(s) selected.