Papers › Extract Free Dense Labels from CLIP

Extract Free Dense Labels from CLIP

2 Dec 2021arXiv:2112.01071archive 2025-07-28

Chong Zhou, Chen Change Loy, Bo Dai

Contrastive Language-Image Pre-training (CLIP) has made a remarkable breakthrough in open-vocabulary zero-shot image recognition. Many recent studies leverage the pre-trained CLIP models for image-level classification and manipulation. In this paper, we wish examine the intrinsic potential of CLIP for pixel-level dense prediction, specifically in semantic segmentation. To this end, with minimal modification, we show that MaskCLIP yields compelling segmentation results on open concepts across various datasets in the absence of annotations and fine-tuning. By adding pseudo labeling and self-training, MaskCLIP+ surpasses SOTA transductive zero-shot semantic segmentation methods by large margins, e.g., mIoUs of unseen classes on PASCAL VOC/PASCAL Context/COCO Stuff are improved from 35.6/20.7/30.3 to 86.1/66.7/54.7. We also test the robustness of MaskCLIP under input corruption and evaluate its capability in discriminating fine-grained objects and novel concepts. Our finding suggests that MaskCLIP can serve as a new reliable source of supervision for dense prediction tasks to achieve annotation-free segmentation. Source code is available at https://github.com/chongzhou96/MaskCLIP.

PaperPDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

chongzhou96/maskclip officialmentioned in papermentioned on GitHubpytorchApache-2.0 report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Novel ConceptsOpen Vocabulary Panoptic SegmentationSegmentationSemantic SegmentationUnsupervised Semantic Segmentation with Language-image Pre-trainingZero Shot SegmentationZero-Shot Semantic Segmentation

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Open Vocabulary Panoptic Segmentation ADE20K MaskCLIP PQ 15.1 #10 of 10 Archive leaderboard report
Semantic Segmentation CC3M-TagMask MaskCLIP mIoU 41.0 #4 of 4 Archive leaderboard report
Unsupervised Semantic Segmentation with Language-image Pre-training ADE20K MaskCLIP Mean IoU (val) 9.8 #12 of 13 Archive leaderboard report
Unsupervised Semantic Segmentation with Language-image Pre-training COCO-Object MaskCLIP mIoU 20.6 #11 of 12 Archive leaderboard report
Unsupervised Semantic Segmentation with Language-image Pre-training COCO-Stuff-171 MaskCLIP mIoU 16.4 #10 of 12 Archive leaderboard report
Unsupervised Semantic Segmentation with Language-image Pre-training COCO-Stuff-27 DenseCLIP mIoU 19.6 #4 of 4 Archive leaderboard report
Unsupervised Semantic Segmentation with Language-image Pre-training COCO-Stuff-27 DenseCLIP pixel accuracy 32.2 #4 of 4 Archive leaderboard report
Unsupervised Semantic Segmentation with Language-image Pre-training Cityscapes val MaskCLIP mIoU 10.0 #12 of 12 Archive leaderboard report
Unsupervised Semantic Segmentation with Language-image Pre-training Cityscapes val MaskCLIP pixel accuracy 35.9 #12 of 12 Archive leaderboard report
Unsupervised Semantic Segmentation with Language-image Pre-training KITTI-STEP DenseCLIP mIoU 15.3 #3 of 3 Archive leaderboard report
Unsupervised Semantic Segmentation with Language-image Pre-training KITTI-STEP DenseCLIP pixel accuracy 34.1 #3 of 3 Archive leaderboard report
Unsupervised Semantic Segmentation with Language-image Pre-training PASCAL Context-59 MaskCLIP mIoU 26.4 #10 of 12 Archive leaderboard report
Unsupervised Semantic Segmentation with Language-image Pre-training PASCAL VOC MaskCLIP mIoU 29.3 #10 of 10 Archive leaderboard report
Unsupervised Semantic Segmentation with Language-image Pre-training PascalVOC-20 MaskCLIP mIoU 74.9 #9 of 10 Archive leaderboard report
Zero Shot Segmentation ADE20K training-free zero-shot segmentation MaskCLIP mIoU 10.2 #5 of 5 Archive leaderboard report
Zero-Shot Semantic Segmentation COCO-Stuff MaskCLIP+ Inductive Setting hIoU - #7 of 15 Archive leaderboard report
Zero-Shot Semantic Segmentation COCO-Stuff MaskCLIP+ Transductive Setting hIoU 45.0 #7 of 15 Archive leaderboard report
Zero-Shot Semantic Segmentation PASCAL VOC MaskCLIP+ Inductive Setting hIoU - #5 of 13 Archive leaderboard report
Zero-Shot Semantic Segmentation PASCAL VOC MaskCLIP+ Transductive Setting hIoU 87.4 #5 of 13 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

CLIP

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections