mlsanityData to 7 Oct 2026

Beyond Prompt Degradation: Prototype-guided Dual-pool Prompting for Incremental Object Detection

arXiv 2603.02286 ↗First posted 2 Mar 2026cs.CV cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

Incremental Object Detection (IOD) aims to continuously learn new object categories without forgetting previously learned ones. Recently, prompt-based methods have gained popularity for their replay-free design and parameter efficiency. However, due to prompt coupling and prompt drift, these methods often suffer from prompt degradation during continual adaptation. To address these issues, we propose a novel prompt-decoupled framework called PDP. PDP innovatively designs a dual-pool prompt decoupling paradigm, which consists of a shared pool used to capture task-general knowledge for forward transfer, and a private pool used to learn task-specific discriminative features. This paradigm explicitly separates task-general and task-specific prompts, preventing interference between prompts and mitigating prompt coupling. In addition, to counteract prompt drift resulting from inconsistent supervision where old foreground objects are treated as background in subsequent tasks, PDP introduces a Prototypical Pseudo-Label Generation (PPG) module. PPG can dynamically update the class prototype space during training and use the class prototypes to further filter valuable pseudo-labels, maintaining supervisory signal consistency throughout the incremental process. PDP achieves state-of-the-art performance on MS-COCO (with a 9.2% AP improvement) and PASCAL VOC (with a 3.3% AP improvement) benchmarks, highlighting its potential in balancing stability and plasticity. The code and dataset are released at: https://github.com/zyt95579/PDPIOD/tree/main

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 43 of its 58 references that are in the map. First posted on arXiv 2 Mar 2026.

Builds on

43 of 58 references.

References · 43

  1. 2014Microsoft COCO: Common Objects in Contextcited by 17,498
  2. 2020Deformable DETR: Deformable Transformers for End-to-End Object Detectioncited by 3,410
  3. 2016Learning without Forgettingcited by 2,465
  4. 2016iCaRL: Incremental Classifier and Representation Learningcited by 2,313
  5. 2020FixMatch: Simplifying Semi-Supervised Learning with Consistency and Confidencecited by 2,080
  6. 2021Learning to Prompt for Continual Learningcited by 938
  7. 2017Lifelong Learning with Dynamically Expandable Networkscited by 689
  8. 2022DualPrompt: Complementary Prompting for Rehearsal-free Continual Learningcited by 635
  9. 2019Gradient based sample selection for online continual learningcited by 568
  10. 2022CODA-Prompt: COntinual Decomposed Attention-based Prompting for Rehearsal-Free Continual Learningcited by 446
  11. 2019Learn to Grow: A Continual Structure Learning Framework for Overcoming Catastrophic Forgettingcited by 302
  12. 2021Towards Open World Object Detectioncited by 297
31 more
  1. 2021Rainbow Memory: Continual Learning with a Memory of Diverse Samplescited by 273
  2. 2022FOSTER: Feature Boosting and Compression for Class-Incremental Learningcited by 228
  3. 2020Mnemonics Training: Multi-Class Incremental Learning without Forgettingcited by 220
  4. 2017Incremental Learning of Object Detectors without Catastrophic Forgettingcited by 208
  5. 2018Reinforced Continual Learningcited by 186
  6. 2019Learning to Remember: A Synaptic Plasticity Driven Framework for Continual Learningcited by 181
  7. 2021OW-DETR: Open-world Detection Transformercited by 129
  8. 2020Memory-Efficient Incremental Learning Through Feature Adaptationcited by 107
  9. 2022Overcoming Catastrophic Forgetting in Incremental Object Detection via Elastic Response Distillationcited by 72
  10. 2020Faster ILOD: Incremental Learning for Object Detectors based on Faster RCNNcited by 61
  11. 2023Continual Detection Transformer for Incremental Object Detectioncited by 59
  12. 2022PROB: Probabilistic Objectness for Open World Object Detectioncited by 57
  13. 2024Consistent Prompting for Rehearsal-Free Continual Learningcited by 52
  14. 2023Prompt-In-Prompt Learning for Universal Image Restorationcited by 43
  15. 2023Augmented Box Replay: Overcoming Foreground Shift for Incremental Object Detectioncited by 32
  16. 2024Convolutional Prompting meets Language Models for Continual Learningcited by 32
  17. 2025Exploring Efficient Open-Vocabulary Segmentation in the Remote Sensingcited by 31
  18. 2022Modeling Missing Annotations for Incremental Learning in Object Detectioncited by 23
  19. 2024Bridge Past and Future: Overcoming Information Asymmetry in Incremental Object Detectioncited by 16
  20. 2024NWPU-MOC: A Benchmark for Fine-grained Multi-category Object Counting in Aerial Imagescited by 14
  21. 2025Rethinking Query-based Transformer for Continual Image Segmentationcited by 14
  22. 2024Decoupled Pseudo-labeling for Semi-Supervised Monocular 3D Object Detectioncited by 11
  23. 2025ProtoDepth: Unsupervised Continual Depth Completion with Prototypescited by 11
  24. 2025UCOD-DPL: Unsupervised Camouflaged Object Detection via Dynamic Pseudo-label Learningcited by 10
  25. 2025RainbowPrompt: Diversity-Enhanced Prompt-Evolving for Continual Learningcited by 10
  26. 2025Revisiting Pool-based Prompt Learning for Few-shot Class-incremental Learningcited by 8
  27. 2024Preventing Catastrophic Forgetting through Memory Networks in Continuous Detectioncited by 7
  28. 2025Scale Efficient Training for Large Datasetscited by 7
  29. 2023Unsupervised Recognition of Unknown Objects for Open-World Object Detectioncited by 6
  30. 2025Demystifying Catastrophic Forgetting in Two-Stage Incremental Object Detectorcited by 5
  31. 2024Not Just Object, But State: Compositional Incremental Learning without Forgettingcited by 2

Built on by

Not cited yet.