</> Code with Papers by Syntology

Papers › Boosting Unsupervised Semantic Segmentation with Principal Mask Proposals

Boosting Unsupervised Semantic Segmentation with Principal Mask Proposals

25 Apr 2024arXiv:2404.16818archive 2025-07-28

Oliver Hahn, Nikita Araslanov, Simone Schaub-Meyer, Stefan Roth

Unsupervised semantic segmentation aims to automatically partition images into semantically meaningful regions by identifying global semantic categories within an image corpus without any form of annotation. Building upon recent advances in self-supervised representation learning, we focus on how to leverage these large pre-trained models for the downstream task of unsupervised segmentation. We present PriMaPs - Principal Mask Proposals - decomposing images into semantically meaningful masks based on their feature representation. This allows us to realize unsupervised semantic segmentation by fitting class prototypes to PriMaPs with a stochastic expectation-maximization algorithm, PriMaPs-EM. Despite its conceptual simplicity, PriMaPs-EM leads to competitive results across various pre-trained backbone models, including DINO and DINOv2, and across different datasets, such as Cityscapes, COCO-Stuff, and Potsdam-3. Importantly, PriMaPs-EM is able to boost results when applied orthogonally to current state-of-the-art unsupervised semantic segmentation pipelines. Code is available at https://github.com/visinf/primaps.

PaperPDFCode

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

Code

visinf/primaps officialmentioned in papermentioned on GitHubpytorch 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

Representation LearningSegmentationSemantic SegmentationUnsupervised Semantic Segmentation

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Unsupervised Semantic Segmentation COCO-Stuff-27 PriMaPs+STEGO (DINO ViT-B/8) Clustering [Accuracy] 57.9 #6 of 29 Archive leaderboard report
Unsupervised Semantic Segmentation COCO-Stuff-27 PriMaPs+STEGO (DINO ViT-B/8) Clustering [mIoU] 29.7 #6 of 29 Archive leaderboard report
Unsupervised Semantic Segmentation COCO-Stuff-27 PriMaPs+HP (DINO ViT-S/8) Clustering [Accuracy] 57.8 #15 of 29 Archive leaderboard report
Unsupervised Semantic Segmentation COCO-Stuff-27 PriMaPs+HP (DINO ViT-S/8) Clustering [mIoU] 25.1 #15 of 29 Archive leaderboard report
Unsupervised Semantic Segmentation Cityscapes test PriMaPs-EM + STEGO (DINO ViT-B/8) Accuracy 78.6 #5 of 14 Archive leaderboard report
Unsupervised Semantic Segmentation Cityscapes test PriMaPs-EM + STEGO (DINO ViT-B/8) mIoU 21.6 #5 of 14 Archive leaderboard report
Unsupervised Semantic Segmentation Cityscapes test PriMaPs-EM (DINO ViT-S/8) Accuracy 81.2 #8 of 14 Archive leaderboard report
Unsupervised Semantic Segmentation Cityscapes test PriMaPs-EM (DINO ViT-S/8) mIoU 19.4 #8 of 14 Archive leaderboard report
Unsupervised Semantic Segmentation Potsdam-3 PriMaPs-EM+HP (DINO ViT-B/8) Accuracy 83.3 #1 of 8 Archive leaderboard report
Unsupervised Semantic Segmentation Potsdam-3 PriMaPs-EM+HP (DINO ViT-B/8) mIoU 71.0 #1 of 8 Archive leaderboard report
Unsupervised Semantic Segmentation Potsdam-3 PriMaPs-EM (DINO ViT-B/8) Accuracy 80.5 #5 of 8 Archive leaderboard report
Unsupervised Semantic Segmentation Potsdam-3 PriMaPs-EM (DINO ViT-B/8) mIoU 67.0 #5 of 8 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

AttentionDINOPCAVision Transformer

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