Papers › EmbedMask: Embedding Coupling for One-stage Instance Segmentation

EmbedMask: Embedding Coupling for One-stage Instance Segmentation

4 Dec 2019arXiv:1912.01954archive 2025-07-28

Hui Ying, Zhaojin Huang, Shu Liu, Tianjia Shao, Kun Zhou

Current instance segmentation methods can be categorized into segmentation-based methods that segment first then do clustering, and proposal-based methods that detect first then predict masks for each instance proposal using repooling. In this work, we propose a one-stage method, named EmbedMask, that unifies both methods by taking advantages of them. Like proposal-based methods, EmbedMask builds on top of detection models making it strong in detection capability. Meanwhile, EmbedMask applies extra embedding modules to generate embeddings for pixels and proposals, where pixel embeddings are guided by proposal embeddings if they belong to the same instance. Through this embedding coupling process, pixels are assigned to the mask of the proposal if their embeddings are similar. The pixel-level clustering enables EmbedMask to generate high-resolution masks without missing details from repooling, and the existence of proposal embedding simplifies and strengthens the clustering procedure to achieve high speed with higher performance than segmentation-based methods. Without any bells and whistles, EmbedMask achieves comparable performance as Mask R-CNN, which is the representative two-stage method, and can produce more detailed masks at a higher speed. Code is available at github.com/yinghdb/EmbedMask.

PaperPDFCode

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

Code

yinghdb/EmbedMask officialmentioned in papermentioned on GitHubpytorch report
TengFeiHan0/Instance-Wise-Depth mentioned on GitHubpytorchNOASSERTION report
gakkiri/EmbedMask mentioned 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

ClusteringInstance SegmentationSegmentationSemantic Segmentation

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Instance Segmentation COCO test-dev EmbedMask (ResNet-101-FPN) AP50 59.1% #89 of 112 Archive leaderboard report
Instance Segmentation COCO test-dev EmbedMask (ResNet-101-FPN) AP75 40.3% #89 of 112 Archive leaderboard report
Instance Segmentation COCO test-dev EmbedMask (ResNet-101-FPN) APL 53% #89 of 112 Archive leaderboard report
Instance Segmentation COCO test-dev EmbedMask (ResNet-101-FPN) APM 40.4% #89 of 112 Archive leaderboard report
Instance Segmentation COCO test-dev EmbedMask (ResNet-101-FPN) APS 17.9% #89 of 112 Archive leaderboard report
Instance Segmentation COCO test-dev EmbedMask (ResNet-101-FPN) mask AP 37.7% #89 of 112 Archive leaderboard report
Instance Segmentation COCO test-dev EmbedMask(R-101-FPN) AP50 59.1 #90 of 112 Archive leaderboard report
Instance Segmentation COCO test-dev EmbedMask(R-101-FPN) AP75 40.3 #90 of 112 Archive leaderboard report
Instance Segmentation COCO test-dev EmbedMask(R-101-FPN) APM 40.4 #90 of 112 Archive leaderboard report
Instance Segmentation COCO test-dev EmbedMask(R-101-FPN) APS 17.9 #90 of 112 Archive leaderboard report
Instance Segmentation COCO test-dev EmbedMask(R-101-FPN) mask AP 37.7 #90 of 112 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

1x1 ConvolutionAverage PoolingBatch NormalizationBottleneck Residual BlockConvolutionGlobal Average PoolingKaiming InitializationMask R-CNNMax PoolingRPNReLUResidual BlockResidual ConnectionRoIAlignSPEEDSoftmax

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