Papers › Panoptic SegFormer: Delving Deeper into Panoptic Segmentation with Transformers

Panoptic SegFormer: Delving Deeper into Panoptic Segmentation with Transformers

8 Sep 2021CVPR 2022 1arXiv:2109.03814archive 2025-07-28

Zhiqi Li, Wenhai Wang, Enze Xie, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo, Tong Lu

Panoptic segmentation involves a combination of joint semantic segmentation and instance segmentation, where image contents are divided into two types: things and stuff. We present Panoptic SegFormer, a general framework for panoptic segmentation with transformers. It contains three innovative components: an efficient deeply-supervised mask decoder, a query decoupling strategy, and an improved post-processing method. We also use Deformable DETR to efficiently process multi-scale features, which is a fast and efficient version of DETR. Specifically, we supervise the attention modules in the mask decoder in a layer-wise manner. This deep supervision strategy lets the attention modules quickly focus on meaningful semantic regions. It improves performance and reduces the number of required training epochs by half compared to Deformable DETR. Our query decoupling strategy decouples the responsibilities of the query set and avoids mutual interference between things and stuff. In addition, our post-processing strategy improves performance without additional costs by jointly considering classification and segmentation qualities to resolve conflicting mask overlaps. Our approach increases the accuracy 6.2\% PQ over the baseline DETR model. Panoptic SegFormer achieves state-of-the-art results on COCO test-dev with 56.2\% PQ. It also shows stronger zero-shot robustness over existing methods. The code is released at \url{https://github.com/zhiqi-li/Panoptic-SegFormer}.

PaperPDFConference PDFCode

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

Code

zhiqi-li/Panoptic-SegFormer officialmentioned in papermentioned on GitHubpytorchApache-2.0 report
claud1234/clft 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

DecoderInstance SegmentationPanoptic SegmentationSegmentation

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Panoptic Segmentation COCO minival Panoptic SegFormer (single-scale) PQ 55.8 #19 of 31 Archive leaderboard report
Panoptic Segmentation COCO minival Panoptic SegFormer (single-scale) PQst 46.9 #19 of 31 Archive leaderboard report
Panoptic Segmentation COCO minival Panoptic SegFormer (single-scale) PQth 61.7 #19 of 31 Archive leaderboard report
Panoptic Segmentation COCO minival Panoptic SegFormer (ResNet-101) PQ 50.6 #23 of 31 Archive leaderboard report
Panoptic Segmentation COCO minival Panoptic SegFormer (ResNet-101) PQst 43.2 #23 of 31 Archive leaderboard report
Panoptic Segmentation COCO minival Panoptic SegFormer (ResNet-101) PQth 55.5 #23 of 31 Archive leaderboard report
Panoptic Segmentation COCO test-dev Panoptic SegFormer (Swin-L) PQ 56.2 #4 of 38 Archive leaderboard report
Panoptic Segmentation COCO test-dev Panoptic SegFormer (Swin-L) PQst 47.0 #4 of 38 Archive leaderboard report
Panoptic Segmentation COCO test-dev Panoptic SegFormer (Swin-L) PQth 62.3 #4 of 38 Archive leaderboard report
Panoptic Segmentation COCO test-dev Panoptic SegFormer (PVTv2-B5) PQ 55.8 #5 of 38 Archive leaderboard report
Panoptic Segmentation COCO test-dev Panoptic SegFormer (PVTv2-B5) PQst 46.5 #5 of 38 Archive leaderboard report
Panoptic Segmentation COCO test-dev Panoptic SegFormer (PVTv2-B5) PQth 61.9 #5 of 38 Archive leaderboard report
Panoptic Segmentation COCO test-dev Panoptic SegFormer (ResNet-101) PQ 50.9 #13 of 38 Archive leaderboard report
Panoptic Segmentation COCO test-dev Panoptic SegFormer (ResNet-101) PQst 43.0 #13 of 38 Archive leaderboard report
Panoptic Segmentation COCO test-dev Panoptic SegFormer (ResNet-101) PQth 56.2 #13 of 38 Archive leaderboard report
Panoptic Segmentation COCO test-dev Panoptic SegFormer (ResNet-50) PQ 50.2 #14 of 38 Archive leaderboard report
Panoptic Segmentation COCO test-dev Panoptic SegFormer (ResNet-50) PQst 42.4 #14 of 38 Archive leaderboard report
Panoptic Segmentation COCO test-dev Panoptic SegFormer (ResNet-50) PQth 55.3 #14 of 38 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

Absolute Position EncodingsAdamAttentionBPEConvolutionDeformable Attention ModuleDeformable DETRDense ConnectionsDetrDropoutFeedforward NetworkLabel SmoothingLayer NormalizationLinear LayerMix-FFNMulti-Head AttentionPosition-Wise Feed-Forward LayerResidual ConnectionSegFormerSoftmaxTransformer

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