mlsanityData to 14 Oct 2026

Vision Backbone Enhancement via Multi-Stage Cross-Scale Attention

arXiv 2308.05872 ↗First posted 10 Aug 2023cs.CVcited by 3

Not on the map. Why?

Abstract (arXiv)

Convolutional neural networks (CNNs) and vision transformers (ViTs) have achieved remarkable success in various vision tasks. However, many architectures do not consider interactions between feature maps from different stages and scales, which may limit their performance. In this work, we propose a simple add-on attention module to overcome these limitations via multi-stage and cross-scale interactions. Specifically, the proposed Multi-Stage Cross-Scale Attention (MSCSA) module takes feature maps from different stages to enable multi-stage interactions and achieves cross-scale interactions by computing self-attention at different scales based on the multi-stage feature maps. Our experiments on several downstream tasks show that MSCSA provides a significant performance boost with modest additional FLOPs and runtime.

Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Builds on 69 of its 88 references that are in the map. First posted on arXiv 10 Aug 2023.

Builds on

69 of 88 references.

References · 69

  1. 2017Attention Is All You Needcited by 60,361
  2. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  3. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,399
  4. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,305
  5. 2014Microsoft COCO: Common Objects in Contextcited by 17,608
  6. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,364
  7. 2021Swin Transformer: Hierarchical Vision Transformer using Shifted Windowscited by 12,139
  8. 2017Mask R-CNNcited by 7,277
  9. 2015Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shiftcited by 7,005
  10. 2016Densely Connected Convolutional Networkscited by 6,720
  11. 2016Layer Normalizationcited by 5,669
  12. 2015Rethinking the Inception Architecture for Computer Visioncited by 5,475
38 more
  1. 2017mixup: Beyond Empirical Risk Minimizationcited by 5,183
  2. 2020Training data-efficient image transformers & distillation through attentioncited by 4,586
  3. 2014Going Deeper with Convolutionscited by 4,241
  4. 2022A ConvNet for the 2020scited by 3,579
  5. 2016DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFscited by 3,241
  6. 2021SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformerscited by 2,940
  7. 2019CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Featurescited by 2,884
  8. 2019Searching for MobileNetV3cited by 1,994
  9. 2021Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction without Convolutionscited by 1,962
  10. 2016Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learningcited by 1,512
  11. 2019Deep High-Resolution Representation Learning for Visual Recognitioncited by 1,420
  12. 2019MMDetection: Open MMLab Detection Toolbox and Benchmarkcited by 1,385
  13. 2017Random Erasing Data Augmentationcited by 1,239
  14. 2014Two-Stream Convolutional Networks for Action Recognition in Videoscited by 1,068
  15. 2016Deep Networks with Stochastic Depthcited by 979
  16. 2021Multiscale Vision Transformerscited by 897
  17. 2021CrossViT: Cross-Attention Multi-Scale Vision Transformer for Image Classificationcited by 806
  18. 2017Focal Loss for Dense Object Detectioncited by 752
  19. 2021PVT v2: Improved Baselines with Pyramid Vision Transformercited by 751
  20. 2018AutoAugment: Learning Augmentation Policies from Datacited by 730
  21. 2019Panoptic Feature Pyramid Networkscited by 645
  22. 2021Twins: Revisiting the Design of Spatial Attention in Vision Transformerscited by 565
  23. 2021CSWin Transformer: A General Vision Transformer Backbone with Cross-Shaped Windowscited by 527
  24. 2021MViTv2: Improved Multiscale Vision Transformers for Classification and Detectioncited by 519
  25. 2020Sparse R-CNN: End-to-End Object Detection with Learnable Proposalscited by 495
  26. 2022MaxViT: Multi-Axis Vision Transformercited by 476
  27. 2021LeViT: a Vision Transformer in ConvNet's Clothing for Faster Inferencecited by 475
  28. 2019Res2Net: A New Multi-scale Backbone Architecturecited by 470
  29. 2019Cascade R-CNN: High Quality Object Detection and Instance Segmentationcited by 429
  30. 2019Learning Deep Transformer Models for Machine Translationcited by 309
  31. 2022Vision Transformer with Deformable Attentioncited by 295
  32. 2021Incorporating Convolution Designs into Visual Transformerscited by 248
  33. 2022Transformers in Medical Imaging: A Surveycited by 245
  34. 2022Neighborhood Attention Transformercited by 239
  35. 2021LocalViT: Analyzing Locality in Vision Transformerscited by 228
  36. 2021Co-Scale Conv-Attentional Image Transformerscited by 224
  37. 2016FractalNet: Ultra-Deep Neural Networks without Residualscited by 216
  38. 2021Multi-Scale Vision Longformer: A New Vision Transformer for High-Resolution Image Encodingcited by 207

50 most cited of 69 All 69 (JSON)

Built on by

Cited by year

Citing papers per year: 2024: 1; 2025: 2.

3 papers.

  1. 2024ICPR 2024 Competition on Multiple Sclerosis Lesion Segmentation -- Methods and Resultscited by 4
  2. 2025Stroke Lesion Segmentation using Multi-Stage Cross-Scale Attentioncited by 0
  3. 2025SurgAnt-ViVQA: Learning to Anticipate Surgical Events through GRU-Driven Temporal Cross-Attentioncited by 0