mlsanityData to 7 Oct 2026

Learning Cross-modality Information Bottleneck Representation for Heterogeneous Person Re-Identification

arXiv 2308.15063 ↗First posted 29 Aug 2023cs.CV cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

Visible-Infrared person re-identification (VI-ReID) is an important and challenging task in intelligent video surveillance. Existing methods mainly focus on learning a shared feature space to reduce the modality discrepancy between visible and infrared modalities, which still leave two problems underexplored: information redundancy and modality complementarity. To this end, properly eliminating the identity-irrelevant information as well as making up for the modality-specific information are critical and remains a challenging endeavor. To tackle the above problems, we present a novel mutual information and modality consensus network, namely CMInfoNet, to extract modality-invariant identity features with the most representative information and reduce the redundancies. The key insight of our method is to find an optimal representation to capture more identity-relevant information and compress the irrelevant parts by optimizing a mutual information bottleneck trade-off. Besides, we propose an automatically search strategy to find the most prominent parts that identify the pedestrians. To eliminate the cross- and intra-modality variations, we also devise a modality consensus module to align the visible and infrared modalities for task-specific guidance. Moreover, the global-local feature representations can also be acquired for key parts discrimination. Experimental results on four benchmarks, i.e., SYSU-MM01, RegDB, Occluded-DukeMTMC, Occluded-REID, Partial-REID and PartialiLIDS dataset, have demonstrated the effectiveness of CMInfoNet.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 60 of its 92 references that are in the map. First posted on arXiv 29 Aug 2023.

Builds on

60 of 92 references.

References · 60

  1. 2017Attention Is All You Needcited by 59,932
  2. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  3. 2016Densely Connected Convolutional Networkscited by 6,691
  4. 2019EfficientNet: Rethinking Model Scaling for Convolutional Neural Networkscited by 6,415
  5. 2022Generative Adversarial Networkscited by 5,603
  6. 2017Squeeze-and-Excitation Networkscited by 4,763
  7. 2018DARTS: Differentiable Architecture Searchcited by 1,782
  8. 2017Random Erasing Data Augmentationcited by 1,235
  9. 2016Deep Variational Information Bottleneckcited by 915
  10. 2018Regularized Evolution for Image Classifier Architecture Searchcited by 902
  11. 2015Deep Learning and the Information Bottleneck Principlecited by 793
  12. 2017In Defense of the Triplet Loss for Person Re-Identificationcited by 784
38 more
  1. 2018MINE: Mutual Information Neural Estimationcited by 680
  2. 2018ProxylessNAS: Direct Neural Architecture Search on Target Task and Hardwarecited by 666
  3. 2020Deep Learning for Person Re-identification: A Survey and Outlookcited by 547
  4. 2018FBNet: Hardware-Aware Efficient ConvNet Design via Differentiable Neural Architecture Searchcited by 480
  5. 2017Person Transfer GAN to Bridge Domain Gap for Person Re-Identificationcited by 479
  6. 2017Beyond Part Models: Person Retrieval with Refined Part Pooling (and a Strong Convolutional Baseline)cited by 472
  7. 2021TransReID: Transformer-based Object Re-Identificationcited by 432
  8. 2016Designing Neural Network Architectures using Reinforcement Learningcited by 277
  9. 2019Searching for A Robust Neural Architecture in Four GPU Hourscited by 258
  10. 2019Progressive Differentiable Architecture Search: Bridging the Depth Gap between Search and Evaluationcited by 230
  11. 2018Harmonious Attention Network for Person Re-Identificationcited by 211
  12. 2019Joint Discriminative and Generative Learning for Person Re-identificationcited by 211
  13. 2017Image-Image Domain Adaptation with Preserved Self-Similarity and Domain-Dissimilarity for Person Re-identificationcited by 168
  14. 2019ABD-Net: Attentive but Diverse Person Re-Identificationcited by 126
  15. 2020Graph Information Bottleneck for Subgraph Recognitioncited by 119
  16. 2021Diverse Part Discovery: Occluded Person Re-identification with Part-Aware Transformercited by 119
  17. 2020Dynamic Dual-Attentive Aggregation Learning for Visible-Infrared Person Re-Identificationcited by 111
  18. 2020Restricting the Flow: Information Bottlenecks for Attributioncited by 110
  19. 2020High-Order Information Matters: Learning Relation and Topology for Occluded Person Re-Identificationcited by 105
  20. 2018Variational Discriminator Bottleneck: Improving Imitation Learning, Inverse RL, and GANs by Constraining Information Flowcited by 103
  21. 2019RGB-Infrared Cross-Modality Person Re-Identification via Joint Pixel and Feature Alignmentcited by 102
  22. 2019Generalization in Reinforcement Learning with Selective Noise Injection and Information Bottleneckcited by 101
  23. 2018Part-Aligned Bilinear Representations for Person Re-identificationcited by 90
  24. 2020Joint Disentangling and Adaptation for Cross-Domain Person Re-Identificationcited by 83
  25. 2020Cross-modality Person re-identification with Shared-Specific Feature Transfercited by 80
  26. 2016A Discriminatively Learned CNN Embedding for Person Re-identificationcited by 78
  27. 2019InfoBot: Transfer and Exploration via the Information Bottleneckcited by 78
  28. 2019Significance-aware Information Bottleneck for Domain Adaptive Semantic Segmentationcited by 74
  29. 2019FasterSeg: Searching for Faster Real-time Semantic Segmentationcited by 73
  30. 2018Occluded Person Re-identificationcited by 72
  31. 2020Cross-Modality Paired-Images Generation for RGB-Infrared Person Re-Identificationcited by 71
  32. 2018FD-GAN: Pose-guided Feature Distilling GAN for Robust Person Re-identificationcited by 70
  33. 2020Pose-guided Visible Part Matching for Occluded Person ReIDcited by 69
  34. 2017Deeply-Learned Part-Aligned Representations for Person Re-Identificationcited by 68
  35. 2017Camera Style Adaptation for Person Re-identificationcited by 67
  36. 2021Learning by Aligning: Visible-Infrared Person Re-identification using Cross-Modal Correspondencescited by 66
  37. 2018Deep Spatial Feature Reconstruction for Partial Person Re-identification: Alignment-Free Approachcited by 58
  38. 2019Foreground-aware Pyramid Reconstruction for Alignment-free Occluded Person Re-identificationcited by 55

50 most cited of 60 All 60 (JSON)

Built on by

Not cited yet.