HyperAI超神経

プラットフォーム

Semantic Segmentation On Lip Val

評価指標

mIoU

評価結果

このベンチマークにおける各モデルのパフォーマンス結果

		Paper Title
Hulk(Finetune, ViT-L)	66.02%	Hulk: A Universal Knowledge Translator for Human-Centric Tasks
Hulk(Finetune, ViT-B)	63.98%	Hulk: A Universal Knowledge Translator for Human-Centric Tasks
UniHCP (finetune)	63.86%	UniHCP: A Unified Model for Human-Centric Perceptions
SOLIDER	60.50%	Beyond Appearance: a Semantic Controllable Self-Supervised Learning Framework for Human-Centric Visual Tasks
HRNetV2 + OCR + RMI (PaddleClas pretrained)	58.2%	Segmentation Transformer: Object-Contextual Representations for Semantic Segmentation
OCR (HRNetV2-W48)	56.65%	Segmentation Transformer: Object-Contextual Representations for Semantic Segmentation
HRNetV2 (HRNetV2-W48)	55.90%	High-Resolution Representations for Labeling Pixels and Regions
OCR (ResNet-101)	55.6%	Segmentation Transformer: Object-Contextual Representations for Semantic Segmentation
CE2P (ResNet-101)	53.10%	Devil in the Details: Towards Accurate Single and Multiple Human Parsing
JPPNet (ResNet-101)	51.37%	Look into Person: Joint Body Parsing & Pose Estimation Network and A New Benchmark
MuLA (ResNet-101)	49.30%	Mutual Learning to Adapt for Joint Human Parsing and Pose Estimation
MMAN (ResNet-101)	46.81%	Macro-Micro Adversarial Network for Human Parsing
Attention+SSL (ResNet-101)	44.73%	Look into Person: Self-supervised Structure-sensitive Learning and A New Benchmark for Human Parsing

0 of 13 row(s) selected.

Semantic Segmentation On Lip Val | SOTA | HyperAI超神経