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

CONDA: Continual Unsupervised Domain Adaptation Learning in Visual Perception for Self-Driving Cars (2212.00621v2)

Published 1 Dec 2022 in cs.CV

Abstract: Although unsupervised domain adaptation methods have achieved remarkable performance in semantic scene segmentation in visual perception for self-driving cars, these approaches remain impractical in real-world use cases. In practice, the segmentation models may encounter new data that have not been seen yet. Also, the previous data training of segmentation models may be inaccessible due to privacy problems. Therefore, to address these problems, in this work, we propose a Continual Unsupervised Domain Adaptation (CONDA) approach that allows the model to continuously learn and adapt with respect to the presence of the new data. Moreover, our proposed approach is designed without the requirement of accessing previous training data. To avoid the catastrophic forgetting problem and maintain the performance of the segmentation models, we present a novel Bijective Maximum Likelihood loss to impose the constraint of predicted segmentation distribution shifts. The experimental results on the benchmark of continual unsupervised domain adaptation have shown the advanced performance of the proposed CONDA method.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (73)
  1. Self-supervised augmentation consistency for adapting semantic segmentation. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2021.
  2. Modeling the background for incremental learning in semantic segmentation. In Proc. Conf. Comp. Vision Pattern Rec., 2020.
  3. Rethinking atrous convolution for semantic image segmentation. arXiv preprint arXiv:1706.05587, 2017a.
  4. Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected CRFs. TPAMI, 2018a.
  5. Encoder-decoder with atrous separable convolution for semantic image segmentation. In ECCV, 2018b.
  6. Domain adaptation for semantic segmentation with maximum squares loss. In ICCV, 2019.
  7. Smoothing matters: Momentum transformer for domain adaptive semantic segmentation. CoRR, 2022.
  8. Road: Reality oriented adaptation for semantic segmentation of urban scenes. In CVPR, 2018c.
  9. No more discrimination: Cross city adaptation of road scene segmenters. In ICCV, 2017b.
  10. The Cityscapes dataset for semantic urban scene understanding. In CVPR, 2016.
  11. Nice: Non-linear independent components estimation, 2015.
  12. Density estimation using real nvp, 2017.
  13. Podnet: Pooled outputs distillation for small-tasks incremental learning. In Proc. Europ. Conf. Computer Vision, 2020.
  14. Plop: Learning without forgetting for continual semantic segmentation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 4040–4050, 2021.
  15. Learning from longitudinal face demonstration—where tractable deep modeling meets inverse reinforcement learning. IJCV, 2019.
  16. Vec2face: Unveil human faces from their blackbox features in face recognition. In CVPR, 2020.
  17. Poda: Prompt-driven zero-shot domain adaptation. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 18623–18633, 2023.
  18. Robert French. Catastrophic forgetting in connectionist networks. 1999.
  19. Unsupervised domain adaptation by backpropagation. In ICML, 2015.
  20. Deep residual learning for image recognition. In CVPR, 2016.
  21. FCNs in the wild: Pixel-level adversarial and constraint-based adaptation. arXiv:1612.02649, 2016.
  22. CyCADA: Cycle-consistent adversarial domain adaptation. In ICML, 2018.
  23. Conditional generative adversarial network for structured domain adaptation. In CVPR, 2018.
  24. DAFormer: Improving network architectures and training strategies for domain-adaptive semantic segmentation. In CVPR, 2022a.
  25. HRDA: Context-aware high-resolution domain-adaptive semantic segmentation. In Proceedings of the European Conference on Computer Vision (ECCV), 2022b.
  26. Mic: Masked image consistency for context-enhanced domain adaptation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 11721–11732, 2023.
  27. Towards open world object detection. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 5830–5840, 2021.
  28. Contrastive adaptation network for unsupervised domain adaptation. In CVPR, 2019.
  29. Glow: Generative flow with invertible 1x1 convolutions. In NIPS, 2018.
  30. Overcoming catastrophic forgetting in neural networks. 2017.
  31. SPIGAN: Privileged adversarial learning from simulation. In ICLR, 2019.
  32. Learning without forgetting. Proc. Europ. Conf. Computer Vision, 2016.
  33. Refinenet: Multi-path refinement networks for high-resolution semantic segmentation. In CVPR, 2017.
  34. Fully convolutional networks for semantic segmentation. In CVPR, 2015a.
  35. Learning transferable features with deep adaptation networks. In ICML, 2015b.
  36. Gradient episodic memory for continual learning. 2017.
  37. Incremental learning techniques for semantic segmentation. 2019a.
  38. Incremental learning techniques for semantic segmentation. In Proceedings of the IEEE/CVF international conference on computer vision workshops, pages 0–0, 2019b.
  39. Image to image translation for domain adaptation. In CVPR, 2018.
  40. The mapillary vistas dataset for semantic understanding of street scenes. In ICCV, 2017.
  41. Temporal non-volume preserving approach to facial age-progression and age-invariant face recognition. In ICCV, 2017.
  42. Extending pretrained segmentation networks with additional anatomical structures. International journal of computer assisted radiology and surgery, 14:1187–1195, 2019.
  43. Learn the new, keep the old: Extending pretrained models with new anatomy and images. In Medical Image Computing and Computer Assisted Intervention–MICCAI 2018: 21st International Conference, Granada, Spain, September 16-20, 2018, Proceedings, Part IV 11, pages 361–369. Springer, 2018.
  44. Unsupervised intra-domain adaptation for semantic segmentation through self-supervision. In CVPR, 2020.
  45. Pytorch: An imperative style, high-performance deep learning library, 2019.
  46. Full-resolution residual networks for semantic segmentation in street scenes. In CVPR, 2017.
  47. Playing for data: Ground truth from computer games. In ECCV, 2016.
  48. Anthony Robins. Catastrophic forgetting, rehearsal and pseudorehearsal. Connection Science, 1995.
  49. The SYNTHIA dataset: A large collection of synthetic images for semantic segmentation of urban scenes. In CVPR, 2016.
  50. Mohammad Rostami. Lifelong domain adaptation via consolidated internal distribution. 2021.
  51. Model adaptation with synthetic and real data for semantic dense foggy scene understanding. In ECCV, 2018.
  52. Multi-target adversarial frameworks for domain adaptation in semantic segmentation. In IEEE/CVF International Conference on Computer Vision (ICCV), 2021.
  53. Multi-head distillation for continual unsupervised domain adaptation in semantic segmentation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshop, 2022.
  54. Sebastian Thrun. Lifelong learning algorithms. In Springer Learning to Learn, 1998.
  55. Bimal: Bijective maximum likelihood approach to domain adaptation in semantic scene segmentation. In International Conference on Computer Vision, 2021a.
  56. Fast flow reconstruction via robust invertible n × n convolution. Future Internet, 2021b.
  57. Fredom: Fairness domain adaptation approach to semantic scene understanding. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2023a.
  58. Fairness continual learning approach to semantic scene understanding in open-world environments. In NeurIPS, 2023b.
  59. Learning to adapt structured output space for semantic segmentation. In CVPR, 2018.
  60. Domain adaptation for structured output via discriminative representations. arXiv:1901.05427, 2019.
  61. Adversarial discriminative domain adaptation. In CVPR, 2017.
  62. Idd: A dataset for exploring problems of autonomous navigation in unconstrained environments. 2019 IEEE Winter Conference on Applications of Computer Vision (WACV), pages 1743–1751, 2019.
  63. Continual adaptation of visual representations via domain randomization and meta-learning. In Proc. Conf. Comp. Vision Pattern Rec., 2021.
  64. Advent: Adversarial entropy minimization for domain adaptation in semantic segmentation. In CVPR, 2019a.
  65. Dada: Depth-aware domain adaptation in semantic segmentation. In ICCV, 2019b.
  66. DCAN: Dual channel-wise alignment networks for unsupervised scene adaptation. In ECCV, 2018.
  67. Segformer: Simple and efficient design for semantic segmentation with transformers. In NeurIPS, 2021.
  68. Multi-scale context aggregation by dilated convolutions. In ICLR, 2016.
  69. Prototypical cross-domain self-supervised learning for few-shot unsupervised domain adaptation. In CVPR, 2021.
  70. Prototypical pseudo label denoising and target structure learning for domain adaptive semantic segmentation. arXiv preprint arXiv:2101.10979, 2021.
  71. Unpaired image-to-image translation using cycle-consistent adversarial networks. In ICCV, 2017.
  72. Penalizing top performers: Conservative loss for semantic segmentation adaptation. In ECCV, 2018.
  73. Unsupervised domain adaptation for semantic segmentation via class-balanced self-training. In ECCV, 2018.
Citations (6)
View on Semantic Scholar

Summary

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

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