mlsanityData to 14 Oct 2026

Expectation-Maximization Attention Networks for Semantic Segmentation

arXiv 1907.13426 ↗First posted 31 Jul 2019cs.CVcited by 147

Not on the map. Why?

Abstract (arXiv)

Self-attention mechanism has been widely used for various tasks. It is designed to compute the representation of each position by a weighted sum of the features at all positions. Thus, it can capture long-range relations for computer vision tasks. However, it is computationally consuming. Since the attention maps are computed w.r.t all other positions. In this paper, we formulate the attention mechanism into an expectation-maximization manner and iteratively estimate a much more compact set of bases upon which the attention maps are computed. By a weighted summation upon these bases, the resulting representation is low-rank and deprecates noisy information from the input. The proposed Expectation-Maximization Attention (EMA) module is robust to the variance of input and is also friendly in memory and computation. Moreover, we set up the bases maintenance and normalization methods to stabilize its training procedure. We conduct extensive experiments on popular semantic segmentation benchmarks including PASCAL VOC, PASCAL Context and COCO Stuff, on which we set new records.

Key facts. Cited by 147 papers in the map (data through 14 Oct 2026). First posted on arXiv 31 Jul 2019.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2021: 50; 2022: 38; 2023: 30; 2024: 17; 2025: 8; 2026: 4.

147 papers.

Sort:

By area:

  1. 2021SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformerscited by 2,940
  2. 2022SegNeXt: Rethinking Convolutional Attention Design for Semantic Segmentationcited by 252
  3. 2021Attention Mechanisms in Computer Vision: A Surveycited by 232
  4. 2022CMX: Cross-Modal Fusion for RGB-X Semantic Segmentation with Transformerscited by 231
  5. 2021K-Net: Towards Unified Image Segmentationcited by 212
  6. 2021Adaptive Prototype Learning and Allocation for Few-Shot Segmentationcited by 207
  7. 2022Rethinking Semantic Segmentation: A Prototype Viewcited by 140
  8. 2021Perturbed and Strict Mean Teachers for Semi-supervised Semantic Segmentationcited by 113
  9. 2021Beyond Self-attention: External Attention using Two Linear Layers for Visual Taskscited by 109
  10. 2022SegViT: Semantic Segmentation with Plain Vision Transformerscited by 106
  11. 2022An Empirical Study of Remote Sensing Pretrainingcited by 105
  12. 2022GMMSeg: Gaussian Mixture based Generative Semantic Segmentation Modelscited by 87
38 more
  1. 2022Deep Hierarchical Semantic Segmentationcited by 84
  2. 2022Learning Affordance Grounding from Exocentric Imagescited by 82
  3. 2021Is Attention Better Than Matrix Decomposition?cited by 77
  4. 2022Representation Compensation Networks for Continual Semantic Segmentationcited by 72
  5. 2021Mining Latent Classes for Few-shot Segmentationcited by 69
  6. 2024Convolution Meets LoRA: Parameter Efficient Finetuning for Segment Anything Modelcited by 69
  7. 2022RbA: Segmenting Unknown Regions Rejected by Allcited by 67
  8. 2022CoCoNet: Coupled Contrastive Learning Network with Multi-level Feature Ensemble for Multi-modality Image Fusioncited by 63
  9. 2023Grounding 3D Object Affordance from 2D Interactions in Imagescited by 62
  10. 2022Learning Non-target Knowledge for Few-shot Semantic Segmentationcited by 57
  11. 2022Expectation-Maximization Contrastive Learning for Compact Video-and-Language Representationscited by 57
  12. 2021SA-Net: Shuffle Attention for Deep Convolutional Neural Networkscited by 54
  13. 2021Capturing Omni-Range Context for Omnidirectional Segmentationcited by 50
  14. 2021Enhanced Boundary Learning for Glass-like Object Segmentationcited by 48
  15. 2021Mining Contextual Information Beyond Image for Semantic Segmentationcited by 48
  16. 2021CTNet: Context-based Tandem Network for Semantic Segmentationcited by 43
  17. 2021Prototypical Cross-Attention Networks for Multiple Object Tracking and Segmentationcited by 43
  18. 2021PVT: Point-Voxel Transformer for Point Cloud Learningcited by 43
  19. 2021PatchFormer: An Efficient Point Transformer with Patch Attentioncited by 43
  20. 2021Active Boundary Loss for Semantic Segmentationcited by 40
  21. 2021One-Shot Object Affordance Detection in the Wildcited by 40
  22. 2021ISNet: Integrate Image-Level and Semantic-Level Context for Semantic Segmentationcited by 40
  23. 2022Visual Attention Methods in Deep Learning: An In-Depth Surveycited by 37
  24. 2021PaddleSeg: A High-Efficient Development Toolkit for Image Segmentationcited by 35
  25. 2022EM-driven unsupervised learning for efficient motion segmentationcited by 33
  26. 2021One-Shot Affordance Detectioncited by 32
  27. 2022Weakly Supervised Temporal Action Localization via Representative Snippet Knowledge Propagationcited by 32
  28. 2022Learning Implicit Feature Alignment Function for Semantic Segmentationcited by 31
  29. 2022StructToken : Rethinking Semantic Segmentation with Structural Priorcited by 30
  30. 2021PointFlow: Flowing Semantics Through Points for Aerial Image Segmentationcited by 29
  31. 2023Making Vision Transformers Efficient from A Token Sparsification Viewcited by 29
  32. 2023Uncovering Prototypical Knowledge for Weakly Open-Vocabulary Semantic Segmentationcited by 27
  33. 2022SWEM: Towards Real-Time Video Object Segmentation with Sequential Weighted Expectation-Maximizationcited by 26
  34. 2022Eliminating Gradient Conflict in Reference-based Line-Art Colorizationcited by 25
  35. 2023SegViTv2: Exploring Efficient and Continual Semantic Segmentation with Plain Vision Transformerscited by 25
  36. 2024Context-Guided Spatial Feature Reconstruction for Efficient Semantic Segmentationcited by 25
  37. 2022Learning Local and Global Temporal Contexts for Video Semantic Segmentationcited by 24
  38. 2023AerialFormer: Multi-resolution Transformer for Aerial Image Segmentationcited by 23

50 most cited of 147 All 147 (JSON)