HyperAI초신경

홈 뉴스 연구 논문 튜토리얼 데이터셋 백과사전 SOTA LLM 모델 GPU 랭킹 컨퍼런스

한국어

HyperAI초신경

Unsupervised Semantic Segmentation With 10

평가 지표

mIoU

평가 결과

이 벤치마크에서 각 모델의 성능 결과

모델 이름	mIoU	Paper Title	Repository
CLS-SEG	35.3	TagCLIP: A Local-to-Global Framework to Enhance Open-Vocabulary Multi-Label Classification of CLIP Without Training
ProxyCLIP	39.2	ProxyCLIP: Proxy Attention Improves CLIP for Open-Vocabulary Segmentation
TagAlign	33.3	TagAlign: Improving Vision-Language Alignment with Multi-Tag Classification
Trident	42.2	Harnessing Vision Foundation Models for High-Performance, Training-Free Open Vocabulary Segmentation
TCL	31.6	Learning to Generate Text-grounded Mask for Open-world Semantic Segmentation from Only Image-Text Pairs
COSMOS ViT-B/16	31.3	COSMOS: Cross-Modality Self-Distillation for Vision Language Pre-training
TTD (TCL)	37.4	TTD: Text-Tag Self-Distillation Enhancing Image-Text Alignment in CLIP to Alleviate Single Tag Bias
TTD (MaskCLIP)	26.5	TTD: Text-Tag Self-Distillation Enhancing Image-Text Alignment in CLIP to Alleviate Single Tag Bias
MaskCLIP	20.6	Extract Free Dense Labels from CLIP
GroupViT (RedCaps)	27.5	GroupViT: Semantic Segmentation Emerges from Text Supervision
ReCo	15.7	ReCo: Retrieve and Co-segment for Zero-shot Transfer

0 of 11 row(s) selected.