AI Paper Atlas

Research map / Benchmarks / PAS-20

PAS-20 leaderboard

10 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on PAS-20. Best mIoU: 98.5 (HyperCLIP++, TPAMI 2026).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

PAS-20 · mIoU

10 papers, each with its best reported number for its own method; higher is better.

#mIoU ↑MethodPaperTopic
198.5HyperCLIP++
CLIP ViT-L/14
HyperCLIP++: Fine-Tuning CLIP for Open-Vocabulary Semantic Segmentation in Hyperbolic Space
TPAMI 2026 · Table 1
CLIP and zero-shot
298.5DPSeg (Inference II)
ConvNeXt-L
DPSeg: Dual-Prompt Cost Volume Learning for Open-Vocabulary Semantic Segmentation
CVPR 2025 · Table 1
CLIP and zero-shot
397.7H-CLIP
CLIP ViT-L/14
Parameter-efficient Fine-tuning in Hyperspherical Space for Open-vocabulary Semantic Segmentation
CVPR 2025 · Table 1
Open-set and few-shot detection
497.5InfoCLIP
CLIP ViT-L/14
InfoCLIP: Bridging Vision-Language Pretraining and Open-Vocabulary Semantic Segmentation via Information-Theoretic Alignment Transfer
AAAI 2026 · Table 1
CLIP and zero-shot
597.3CP-CLIP
CLIP ViT-L/14
CP-CLIP: Customized Parameter Generation for Open-vocabulary Semantic Segmentation
AAAI 2026 · Table 1
Open-set and few-shot detection
697.3PCA-Seg
ViT-B/16
PCA-Seg: Revisiting Cost Aggregation for Open-Vocabulary Semantic and Part Segmentation
CVPR 2026 · Table 1
Open-set and few-shot detection
797CAT-Seg
CLIP ViT-L/14
CAT-Seg: Cost Aggregation for Open-Vocabulary Semantic Segmentation
CVPR 2024 · Table 1
CLIP and zero-shot
896.1SED
ConvNeXt-L
SED++: A Simple Encoder-Decoder for Improved Open-Vocabulary Semantic Segmentation
TPAMI 2025 · Table 1
Open-set and few-shot detection
996.1SED
ConvNeXt-L
SED: A Simple Encoder-Decoder for Open-Vocabulary Semantic Segmentation
CVPR 2024 · Table 1
CNN features
1095.4LoGoSeg
CLIP ViT-B/16
LoGoSeg: Integrating Local and Global Features for Open-Vocabulary Semantic Segmentation
AAAI 2026 · Table 1
Open-set and few-shot detection

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.