Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
97 tokens/sec
GPT-4o
53 tokens/sec
Gemini 2.5 Pro Pro
44 tokens/sec
o3 Pro
5 tokens/sec
GPT-4.1 Pro
47 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Periodically Exchange Teacher-Student for Source-Free Object Detection (2311.13930v1)

Published 23 Nov 2023 in cs.CV

Abstract: Source-free object detection (SFOD) aims to adapt the source detector to unlabeled target domain data in the absence of source domain data. Most SFOD methods follow the same self-training paradigm using mean-teacher (MT) framework where the student model is guided by only one single teacher model. However, such paradigm can easily fall into a training instability problem that when the teacher model collapses uncontrollably due to the domain shift, the student model also suffers drastic performance degradation. To address this issue, we propose the Periodically Exchange Teacher-Student (PETS) method, a simple yet novel approach that introduces a multiple-teacher framework consisting of a static teacher, a dynamic teacher, and a student model. During the training phase, we periodically exchange the weights between the static teacher and the student model. Then, we update the dynamic teacher using the moving average of the student model that has already been exchanged by the static teacher. In this way, the dynamic teacher can integrate knowledge from past periods, effectively reducing error accumulation and enabling a more stable training process within the MT-based framework. Further, we develop a consensus mechanism to merge the predictions of two teacher models to provide higher-quality pseudo labels for student model. Extensive experiments on multiple SFOD benchmarks show that the proposed method achieves state-of-the-art performance compared with other related methods, demonstrating the effectiveness and superiority of our method on SFOD task.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (52)
  1. Exploring object relation in mean teacher for cross-domain detection. In CVPR, pages 11457–11466, 2019.
  2. Harmonizing transferability and discriminability for adapting object detectors. In CVPR, pages 8869–8878, 2020.
  3. Reusing the task-specific classifier as a discriminator: Discriminator-free adversarial domain adaptation. In CVPR, pages 7181–7190, 2022.
  4. Diffusiondet: Diffusion model for object detection. arXiv preprint arXiv:2211.09788, 2022.
  5. Self-supervised noisy label learning for source-free unsupervised domain adaptation. In IROS, pages 10185–10192, 2022.
  6. Unsupervised image classification for deep representation learning. In ECCVw, pages 430–446. Springer, 2020.
  7. Domain adaptive faster r-cnn for object detection in the wild. In CVPR, pages 3339–3348, 2018.
  8. Adversarial alignment for source free object detection. In Brian Williams, Yiling Chen, and Jennifer Neville, editors, AAAI, pages 452–460, 2023.
  9. The cityscapes dataset for semantic urban scene understanding. In CVPR, pages 3213–3223, 2016.
  10. Unbiased mean teacher for cross-domain object detection. In CVPR, pages 4091–4101, 2021.
  11. Domain-adversarial training of neural networks. J. Mach. Learn. Res., 17(1):2096–2030, 2016.
  12. Vision meets robotics: The kitti dataset. Int. J. Robotics Res., 32(11):1231–1237, 2013.
  13. Multi-adversarial faster-rcnn for unrestricted object detection. In ICCV, pages 6668–6677, 2019.
  14. Domain adaptive object detection via asymmetric tri-way faster-rcnn. In ECCV 2020, pages 309–324. Springer, 2020.
  15. Source-free unsupervised domain adaptation for cross-modality abdominal multi-organ segmentation. Knowl. Based Syst., 250:109155, 2022.
  16. Progressive domain adaptation for object detection. In WACV, pages 749–757, 2020.
  17. Model adaptation: Historical contrastive learning for unsupervised domain adaptation without source data. NIPS, 34:3635–3649, 2021.
  18. Driving in the matrix: Can virtual worlds replace human-generated annotations for real world tasks? In ICRA, pages 746–753, 2017.
  19. A robust learning approach to domain adaptive object detection. In ICCV, pages 480–490, 2019.
  20. Diversify and match: A domain adaptive representation learning paradigm for object detection. In CVPR, pages 12456–12465, 2019.
  21. Spatial attention pyramid network for unsupervised domain adaptation. In ECCV, pages 481–497. Springer, 2020.
  22. Yolov6: A single-stage object detection framework for industrial applications. arXiv preprint arXiv:2209.02976, 2022.
  23. Divergence-agnostic unsupervised domain adaptation by adversarial attacks. IEEE Trans. Patt. Anal. Mach. Intell., 44(11):8196–8211, 2021.
  24. Source-free object detection by learning to overlook domain style. In CVPR, pages 8014–8023, 2022.
  25. A free lunch for unsupervised domain adaptive object detection without source data. In AAAI, volume 35, pages 8474–8481, 2021.
  26. Target-aware auto-augmentation for unsupervised domain adaptive object detection. In ICASSP, pages 3848–3852, 2022.
  27. Do we really need to access the source data? source hypothesis transfer for unsupervised domain adaptation. In ICML, pages 6028–6039, 2020.
  28. Run and chase: Towards accurate source-free domain adaptive object detection. In ICME, 2023.
  29. Slimmable domain adaptation. In CVPR, pages 7141–7150, 2022.
  30. Simrod: A simple adaptation method for robust object detection. In ICCV, pages 3570–3579, 2021.
  31. Faster r-cnn: Towards real-time object detection with region proposal networks. NIPS, 28, 2015.
  32. Strong-weak distribution alignment for adaptive object detection. In CVPR, pages 6956–6965, 2019.
  33. Semantic foggy scene understanding with synthetic data. IJCV, 126(9):973–992, 2018.
  34. Weighted boxes fusion: Ensembling boxes from different object detection models. Image Vis. Comput., 107:104117, 2021.
  35. Prior knowledge guided unsupervised domain adaptation. In ECCV, pages 639–655, 2022.
  36. Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results. NIPS, 30, 2017.
  37. Adversarial discriminative domain adaptation. In CVPR, pages 7167–7176, 2017.
  38. Mega-cda: Memory guided attention for category-aware unsupervised domain adaptive object detection. In CVPR, pages 4516–4526, 2021.
  39. Instance relation graph guided source-free domain adaptive object detection. In CVPR, pages 3520–3530, 2023.
  40. Mixture of teacher experts for source-free domain adaptive object detection. In ICIP, pages 3606–3610, 2022.
  41. Afan: Augmented feature alignment network for cross-domain object detection. IEEE Trans. Image Process., 30:4046–4056, 2021.
  42. Detectron2. https://github.com/facebookresearch/detectron2, 2019.
  43. Source data-free domain adaptation of object detector through domain-specific perturbation. Int. J. Intell. Syst., 36(8):3746–3766, 2021.
  44. Exploring categorical regularization for domain adaptive object detection. In CVPR, pages 11724–11733, 2020.
  45. Source free domain adaptation for medical image segmentation with fourier style mining. Medical Image Anal., 79:102457, 2022.
  46. Tvt: Transferable vision transformer for unsupervised domain adaptation. In WACV, pages 520–530, 2023.
  47. Bdd100k: A diverse driving video database with scalable annotation tooling. arXiv preprint arXiv:1805.04687, 2(5):6, 2018.
  48. Simulation-and-mining: Towards accurate source-free unsupervised domain adaptive object detection. In ICASSP, pages 3843–3847, 2022.
  49. Self-guided adaptation: Progressive representation alignment for domain adaptive object detection. IEEE Trans. Multim., 24:2246–2258, 2021.
  50. Collaborative training between region proposal localization and classification for domain adaptive object detection. In ECCV, pages 86–102, 2020.
  51. Adaptive mutual learning for unsupervised domain adaptation. IEEE Trans. Circuits Syst. Video Technol., 2023.
  52. ifan: Image-instance full alignment networks for adaptive object detection. In AAAI, volume 34, pages 13122–13129, 2020.
User Edit Pencil Streamline Icon: https://streamlinehq.com
Authors (4)
  1. Qipeng Liu (24 papers)
  2. Luojun Lin (13 papers)
  3. Zhifeng Shen (5 papers)
  4. Zhifeng Yang (3 papers)
Citations (13)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now