• Corpus ID: 226227144

Pixel-Level Cycle Association: A New Perspective for Domain Adaptive Semantic Segmentation

@article{Kang2020PixelLevelCA,
  title={Pixel-Level Cycle Association: A New Perspective for Domain Adaptive Semantic Segmentation},
  author={Guoliang Kang and Yunchao Wei and Yi Yang and Yueting Zhuang and Alexander G. Hauptmann},
  journal={ArXiv},
  year={2020},
  volume={abs/2011.00147}
}
Domain adaptive semantic segmentation aims to train a model performing satisfactory pixel-level predictions on the target with only out-of-domain (source) annotations. The conventional solution to this task is to minimize the discrepancy between source and target to enable effective knowledge transfer. Previous domain discrepancy minimization methods are mainly based on the adversarial training. They tend to consider the domain discrepancy globally, which ignore the pixel-wise relationships and… 

Figures and Tables from this paper

Semantic Distribution-aware Contrastive Adaptation for Semantic Segmentation

TLDR
A new semantic distribution-aware contrastive adaptation algorithm, dubbed as SDCA, that enables pixel-wise representation alignment across domains under the guidance of the semantic distributions, and empirically unveil the certain mechanisms that promote the potential of SDCA.

PixMatch: Unsupervised Domain Adaptation via Pixelwise Consistency Training

TLDR
This work presents a novel framework for unsupervised domain adaptation based on the notion of target-domain consistency training, and introduces a new loss term to enforce pixelwise consistency between the model’s predictions on a target image and a perturbed version of the same image.

SePiCo: Semantic-Guided Pixel Contrast for Domain Adaptive Semantic Segmentation

TLDR
This work proposes SePiCo, a novel one-stage adaptation framework that highlights the semantic concepts of individual pixel to promote learning of class-discriminative and class-balanced pixel embedding space across domains, eventually boosting the performance of self-training methods.

Domain Adaptive Semantic Segmentation without Source Data: Align, Teach and Propagate

TLDR
A unified framework called ATP is proposed, which consists of three schemes, i.e., feature Alignment, bidirectional Teaching, and information Propagation, which devise a curriculum-style entropy minimization objective to implicitly align the target features with unseen source features via the provided source model.

Contrastive Learning and Self-Training for Unsupervised Domain Adaptation in Semantic Segmentation

TLDR
This work proposes a contrastive learning approach that adapts category-wise centroids across domains and extends it with selftraining, where it uses a memory-efficient temporal ensemble to generate consistent and reliable pseudo-labels.

DecoupleNet: Decoupled Network for Domain Adaptive Semantic Segmentation

TLDR
This work proposes DecoupleNet that alleviates source domain overfitting and enables the final model to focus more on the segmentation task, and puts forward Self-Discrimination (SD) and introduces an auxiliary classifier to learn more discriminative target domain features with pseudo labels.

Give Me Your Trained Model: Domain Adaptive Semantic Segmentation without Source Data

TLDR
This work focuses on an interesting and challenging cross-domain semantic segmentation task where only the trained source model is provided to the target domain, and proposes a unified framework called Domain Adaptive Semantic Segmentation without Source data (DAS for short).

SPCL: A New Framework for Domain Adaptive Semantic Segmentation via Semantic Prototype-based Contrastive Learning

TLDR
A novel semantic prototype-based contrastive learning framework for fine-grained class alignment with an elegant contrastive loss to enforce the network to approximate the positive-concentrated and negative-separated prop-erties of cross-domain pixel representations.

Dual Path Learning for Domain Adaptation of Semantic Segmentation

TLDR
This paper proposes a novel dual path learning (DPL) framework that contains two complementary and interactive single-domain adaptation pipelines aligned in source and target domain respectively and shows superiority of the DPL model over the state-of-the-art methods.

Dual Bipartite Graph Learning: A General Approach for Domain Adaptive Object Detection

TLDR
This paper presents a new and general perspective for DAOD named Dual Bipartite Graph Learning (DBGL), which captures the cross-domain interactions on both pixel-level and semantic-level via increasing the distinction between foregrounds and backgrounds and modeling theCross-domain dependencies among different semantic categories.
...

References

SHOWING 1-10 OF 55 REFERENCES

Content-Consistent Matching for Domain Adaptive Semantic Segmentation

TLDR
This paper considers the adaptation of semantic segmentation from the synthetic source domain to the real target domain and proposes a content-consistent matching (CCM) model, which yields consistent improvements over the baselines and performs favorably against previous state-of-the-arts tasks.

SSF-DAN: Separated Semantic Feature Based Domain Adaptation Network for Semantic Segmentation

TLDR
A Separated Semantic Feature based domain adaptation network, named SSF-DAN, for semantic segmentation, designed to independently adapt semantic features across the target and source domains and demonstrates robust performance, superior to state-of-the-art methods on benchmark datasets.

Learning to Adapt Structured Output Space for Semantic Segmentation

TLDR
A multi-level adversarial network is constructed to effectively perform output space domain adaptation at different feature levels and it is shown that the proposed method performs favorably against the state-of-the-art methods in terms of accuracy and visual quality.

CrDoCo: Pixel-Level Domain Transfer With Cross-Domain Consistency

TLDR
A novel pixel-wise adversarial domain adaptation algorithm that leverages image-to-image translation methods for data augmentation and introduces a cross-domain consistency loss that enforces the adapted model to produce consistent predictions.

Unsupervised Intra-Domain Adaptation for Semantic Segmentation Through Self-Supervision

TLDR
This work proposes a two-step self-supervised domain adaptation approach to minimize the inter-domain and intra-domain gap together and proposes to employ a self- supervised adaptation technique from the easy to the hard subdomain.

Domain Adaptation for Structured Output via Discriminative Patch Representations

TLDR
A domain adaptation method to adapt the source data to the unlabeled target domain by discovering multiple modes of patch-wise output distribution through the construction of a clustered space and using an adversarial learning scheme to push the feature representations of target patches in the clustered space closer to the distributions of source patches.

ADVENT: Adversarial Entropy Minimization for Domain Adaptation in Semantic Segmentation

TLDR
This work proposes two novel, complementary methods using (i) entropy loss and (ii) adversarial loss respectively for unsupervised domain adaptation in semantic segmentation with losses based on the entropy of the pixel-wise predictions.

Constructing Self-Motivated Pyramid Curriculums for Cross-Domain Semantic Segmentation: A Non-Adversarial Approach

TLDR
This work proposes a new approach, called self-motivated pyramid curriculum domain adaptation (PyCDA), to facilitate the adaptation of semantic segmentation neural networks from synthetic source domains to real target domains and reports state-of-the-art results for the adaptation from both GTAV and SYNTHIA to Cityscapes.

Regularizing Proxies with Multi-Adversarial Training for Unsupervised Domain-Adaptive Semantic Segmentation

TLDR
This work proposes a novel proxy-based method using multi-adversarial training to generate high-quality labels as proxy labels to supervise the training on target data and achieves state-of-the-art results on two major tasks.

Domain Adaptation for Semantic Segmentation With Maximum Squares Loss

TLDR
The maximum squares loss prevents the training process being dominated by easy-to-transfer samples in the target domain, and introduces the image-wise weighting ratio to alleviate the class imbalance in the unlabeled target domain.
...