mlsanityData to 7 Oct 2026

Socialized Detector Learning: Trajectory-Guided and Reciprocal Distillation for Heterogeneous Object Detectors

arXiv 2608.25836 ↗First posted 26 Aug 2026cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

Object detection knowledge is fragmented across independently trained, heterogeneous detectors with complementary category supports. In socialized learning, this knowledge resides in a society, and learning aims to evolve the society collectively through exchange. However, aggregation-based socialization does not explicitly plan transfer order, whereas progressive multi-teacher distillation considers order but remains a one-way student enhancement in a shared category space. Building on Socialized Learning, we formulate Socialized Detector Learning (SDL) for heterogeneous, category-specialized object detectors and propose Trajectory-Guided and Reciprocal Distillation (TGRD).TGRD estimates directed operational Inter-Detector Transfer Difficulty (IDTD) from held-out feature-alignment residuals, precomputes a fixed score table, and greedily constructs a carrier trajectory. Along the trajectory, knowledge is progressively consolidated into a union-category carrier and then returned to experts through reciprocal transfer. A conditional proxy-certificate analysis shows that, under stated assumptions, the progressive certificate is no larger than an aggregated-target counterpart. On MS COCO with four heterogeneous experts and two carrier initializations, final carriers outperform epoch-matched simultaneous aggregation controls by 2.6 AP in both settings. Reciprocal detectors attain 20.8--28.4 AP on previously unsupported categories while remaining within 1.3 AP of original expert-specific performance. These results support order-aware progressive consolidation followed by reciprocal transfer as a viable mechanism for detector-society evolution.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 19 of its 33 references that are in the map. First posted on arXiv 26 Aug 2026.

Builds on

19 of 33 references.

References · 19

  1. 2014Microsoft COCO: Common Objects in Contextcited by 17,498
  2. 2015Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networkscited by 9,617
  3. 2019FCOS: Fully Convolutional One-Stage Object Detectioncited by 1,573
  4. 2020Ensemble Distillation for Robust Model Fusion in Federated Learningcited by 678
  5. 2020Federated Learning with Matched Averagingcited by 564
  6. 2020Generalized Focal Loss: Learning Qualified and Distributed Bounding Boxes for Dense Object Detectioncited by 408
  7. 2017Incremental Learning of Object Detectors without Catastrophic Forgettingcited by 208
  8. 2019Distilling Object Detectors with Fine-grained Feature Imitationcited by 188
  9. 2020FedBE: Making Bayesian Model Ensemble Applicable to Federated Learningcited by 176
  10. 2021Focal and Global Knowledge Distillation for Detectorscited by 138
  11. 2021General Instance Distillation for Object Detectioncited by 113
  12. 2021Distilling Object Detectors via Decoupled Featurescited by 105
7 more
  1. 2022Overcoming Catastrophic Forgetting in Incremental Object Detection via Elastic Response Distillationcited by 72
  2. 2021Localization Distillation for Dense Object Detectioncited by 55
  3. 2023CrossKD: Cross-Head Knowledge Distillation for Object Detectioncited by 50
  4. 2022DETRDistill: A Universal Knowledge Distillation Framework for DETR-familiescited by 36
  5. 2020SID: Incremental Learning for Anchor-Free Object Detection via Selective and Inter-Related Distillationcited by 32
  6. 2023FedLPA: One-shot Federated Learning with Layer-Wise Posterior Aggregationcited by 22
  7. 2024Bridge Past and Future: Overcoming Information Asymmetry in Incremental Object Detectioncited by 16

Built on by

Not cited yet.