mlsanityData to 14 Oct 2026

Copy-Pasting Coherent Depth Regions Improves Contrastive Learning for Urban-Scene Segmentation

arXiv 2211.14074 ↗First posted 25 Nov 2022cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

In this work, we leverage estimated depth to boost self-supervised contrastive learning for segmentation of urban scenes, where unlabeled videos are readily available for training self-supervised depth estimation. We argue that the semantics of a coherent group of pixels in 3D space is self-contained and invariant to the contexts in which they appear. We group coherent, semantically related pixels into coherent depth regions given their estimated depth and use copy-paste to synthetically vary their contexts. In this way, cross-context correspondences are built in contrastive learning and a context-invariant representation is learned. For unsupervised semantic segmentation of urban scenes, our method surpasses the previous state-of-the-art baseline by +7.14% in mIoU on Cityscapes and +6.65% on KITTI. For fine-tuning on Cityscapes and KITTI segmentation, our method is competitive with existing models, yet, we do not need to pre-train on ImageNet or COCO, and we are also more computationally efficient. Our code is available on https://github.com/LeungTsang/CPCDR

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 47 of its 59 references that are in the map. First posted on arXiv 25 Nov 2022.

Builds on

47 of 59 references.

References · 47

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2014Microsoft COCO: Common Objects in Contextcited by 17,608
  3. 2020A Simple Framework for Contrastive Learning of Visual Representationscited by 12,049
  4. 2015Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networkscited by 9,639
  5. 2014ImageNet Large Scale Visual Recognition Challengecited by 8,498
  6. 2019Momentum Contrast for Unsupervised Visual Representation Learningcited by 7,513
  7. 2021Masked Autoencoders Are Scalable Vision Learnerscited by 7,403
  8. 2017Mask R-CNNcited by 7,277
  9. 2014Fully Convolutional Networks for Semantic Segmentationcited by 4,859
  10. 2016The Cityscapes Dataset for Semantic Urban Scene Understandingcited by 4,444
  11. 2019CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Featurescited by 2,884
  12. 2020Exploring Simple Siamese Representation Learningcited by 2,866
35 more
  1. 2020Unsupervised Learning of Visual Features by Contrasting Cluster Assignmentscited by 2,857
  2. 2013Sinkhorn Distances: Lightspeed Computation of Optimal Transportation Distancescited by 2,316
  3. 2020Improved Baselines with Momentum Contrastive Learningcited by 2,287
  4. 2020Shortcut Learning in Deep Neural Networkscited by 1,969
  5. 2021Barlow Twins: Self-Supervised Learning via Redundancy Reductioncited by 1,838
  6. 2021An Empirical Study of Training Self-Supervised Vision Transformerscited by 1,616
  7. 2018Unsupervised Representation Learning by Predicting Image Rotationscited by 1,591
  8. 2016Unsupervised Learning of Visual Representations by Solving Jigsaw Puzzlescited by 1,382
  9. 2020Big Self-Supervised Models are Strong Semi-Supervised Learnerscited by 1,380
  10. 2016Colorful Image Colorizationcited by 1,114
  11. 2015Unsupervised Visual Representation Learning by Context Predictioncited by 1,066
  12. 2020What Makes for Good Views for Contrastive Learning?cited by 846
  13. 2017Unsupervised Learning of Depth and Ego-Motion from Videocited by 782
  14. 2019Panoptic Feature Pyramid Networkscited by 645
  15. 2020Simple Copy-Paste is a Strong Data Augmentation Method for Instance Segmentationcited by 596
  16. 2020Dense Contrastive Learning for Self-Supervised Visual Pre-Trainingcited by 512
  17. 2020Propagate Yourself: Exploring Pixel-Level Consistency for Unsupervised Visual Representation Learningcited by 297
  18. 2021Unsupervised Semantic Segmentation by Contrasting Object Mask Proposalscited by 190
  19. 2018Tracking Emerges by Colorizing Videoscited by 166
  20. 2021PiCIE: Unsupervised Semantic Segmentation using Invariance and Equivariance in Clusteringcited by 163
  21. 2021Efficient Visual Pretraining with Contrastive Detectioncited by 150
  22. 2020Unsupervised Learning of Dense Visual Representationscited by 147
  23. 2016Learning Features by Watching Objects Movecited by 138
  24. 2019InstaBoost: Boosting Instance Segmentation via Probability Map Guided Copy-Pastingcited by 95
  25. 2021Unsupervised Object-Level Representation Learning from Scene Imagescited by 72
  26. 2020Monocular Depth Estimation Based On Deep Learning: An Overviewcited by 68
  27. 2020Three Ways to Improve Semantic Segmentation with Self-Supervised Depth Estimationcited by 66
  28. 2020Self-Supervised Visual Representation Learning from Hierarchical Groupingcited by 63
  29. 2020CASTing Your Model: Learning to Localize Improves Self-Supervised Representationscited by 63
  30. 2020Distilling Localization for Self-Supervised Representation Learningcited by 34
  31. 2019Putting visual object recognition in contextcited by 33
  32. 2022CP2: Copy-Paste Contrastive Pretraining for Semantic Segmentationcited by 26
  33. 2021RegionCL: Can Simple Region Swapping Contribute to Contrastive Learning?cited by 11
  34. 2017Exploring the Bounds of the Utility of Context for Object Detectioncited by 10
  35. 2022Evaluating Context for Deep Object Detectorscited by 4

Built on by

Not cited yet.