Skip to content

Latest commit

 

History

History
364 lines (195 loc) · 43.2 KB

Domain_Adaptation.md

File metadata and controls

364 lines (195 loc) · 43.2 KB
Raw

Contents for Domain Adaptation

1) Defination

Domain Adaptation belongs to Semi-supervised or Un-supervised Learning / Transfer Learning / Few-shot Learning. We especially focus on domain adaptative object detection for building robust object detection methods in real application.

2) Pioneers and Experts

[Mingsheng Long]

-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-

3) Datasets

-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-

4) Materials

-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-|-+-

5) Papers

*********************************

① ⭐⭐⭐Domain Adaptation for Image Classification

  • ❤Model Evaluation(CVPR2021) Are Labels Necessary for Classifier Accuracy Evaluation?(测试集没有标签，可以拿来测试模型吗？) [arxiv link][CSDN blog]

  • ❤PCS-FUDA(CVPR2021) Prototypical Cross-domain Self-supervised Learning for Few-shot Unsupervised Domain Adaptation [arxiv link][project link][codes|official PyTorch]

  • ❤SHOT++(TPAMI2021) Source Data-Absent Unsupervised Domain Adaptation Through Hypothesis Transfer and Labeling Transfer [paper link][codes|official]

  • PTMDA(TIP2022) Multi-Source Unsupervised Domain Adaptation via Pseudo Target Domain [paper link]

  • DINE(CVPR2022) DINE: Domain Adaptation From Single and Multiple Black-Box Predictors [paper link][codes|official]

*********************************

② ⭐⭐Domain Adaptation for Object Detection

  • ❤DA-FasterRCNN(CVPR2018)(Baseline & Milestone) Domain Adaptive Faster R-CNN for Object Detection in the Wild [arxiv link][paper link][codes|official Caffe][Zhihu blog]

  • ❤cross_domain_detection(CVPR2018) Cross-Domain Weakly-Supervised Object Detection Through Progressive Domain Adaptation [paper link][arxiv link][project link][code|official][Weakly-Supervised Learning for Object Detection]

  • SCL(arxiv2019) SCL: Towards Accurate Domain Adaptive Object Detection via Gradient Detach Based Stacked Complementary Losses [paper link] [code|official]

  • MAF(ICCV2019) Multi-adversarial Faster-RCNN for Unrestricted Object Detection [paper link][No code]

  • DM(CVPR2019) Diversify and Match: A Domain Adaptive Representation Learning Paradigm for Object Detection [paper link]

  • Strong-Weak DA(CVPR2019) Strong-Weak Distribution Alignment for Adaptive Object Detection [arxiv link][project link][codes|official PyTorch]

  • MEAA(ACMMM2020) Domain-Adaptive Object Detection via Uncertainty-Aware Distribution Alignment [paper link][No code]

  • (ECCV2020) YOLO in the Dark: Domain Adaptation Method for Merging Multiple Models [paper link][No code]

  • ATF(ECCV2020) Domain Adaptive Object Detection via Asymmetric Tri-Way Faster-RCNN [paper link][No code]

  • DA-FCOS(ECCV2020) One-Shot Unsupervised Cross-Domain Detection [paper link]

  • CDRA(CVPR2020) Exploring Categorical Regularization for Domain Adaptive Object Detection[paper link][code|official]

  • HTCN(CVPR2020) Harmonizing Transferability and Discriminability for Adapting Object Detectors [paper link][codes|official PyTorch][CSDN blog]

  • PA-ATF(TCSVT2021) Partial Alignment for Object Detection in the Wild [paper link][No code]

  • ❤Divide-and-Merge Spindle Network(DMSN)(ICCV2021) Multi-Source Domain Adaptation for Object Detection [paper link]

  • ❤UMT(CVPR2021) Unbiased Mean Teacher for Cross-domain Object Detection [arxiv link][paper link][codes|official PyTorch]

  • Survey(arxiv2021) Unsupervised Domain Adaptation of Object Detectors: A Survey [paper link]

  • MS-DAYOLO(ICIP2021)(YOLOV4) Multiscale Domain Adaptive YOLO for Cross-Domain Object Detection [arxiv link][csdn blog]

  • DAYOLO(ACML2021)(YOLOV3) Domain Adaptive YOLO for One-Stage Cross-Domain Detection [paper link]

  • US-DAF(ACMMM2022) Universal Domain Adaptive Object Detector [paper link][No code]

  • SCAN(AAAI2022) SCAN: Cross Domain Object Detection with Semantic Conditioned Adaptation [paper link][codes|official PyTorch]

  • SIGMA(CVPR2022) SIGMA: Semantic-complete Graph Matching for Domain Adaptive Object Detection [paper link][codes|official PyTorch]

  • TIA(CVPR2022) Task-specific Inconsistency Alignment for Domain Adaptive Object Detection [paper link][codes|official PyTorch]

  • TPKP(CVPR2022) Target-Relevant Knowledge Preservation for Multi-Source Domain Adaptive Object Detection [paper link][codes|(not found)]

  • MGADA(CVPR2022) Multi-Granularity Alignment Domain Adaptation for Object Detection [paper link][codes|(not found)][related journal link]

  • TDD(CVPR2022) Cross Domain Object Detection by Target-Perceived Dual Branch Distillation [paper link][codes|official PyTorch]

  • AT(CVPR2022) Cross-Domain Adaptive Teacher for Object Detection [paper link][No code]

  • ❤PT(ICML2022) Learning Domain Adaptive Object Detection with Probabilistic Teacher [paper link][code|official][Probabilistic Teacher, Knowledge Distillation Framework]

  • DICN(TPAMI2022) Dual Instance-Consistent Network for Cross-Domain Object Detection [paper link]

  • DenseTeacher(ECCV2022) DenseTeacher: Dense Pseudo-Label for Semi-supervised Object Detection [paper link][code|official]

  • SSDA-YOLO(CVIU2023) SSDA-YOLO: Semi-supervised Domain Adaptive YOLO for Cross-Domain Object Detection [paper link][arxiv link][code|official]

  • DETR-GA(CVPR2023) DETR with Additional Global Aggregation for Cross-domain Weakly Supervised Object Detection [paper link][cross-domain weakly supervised object detection (CDWSOD)]

  • 2PCNet(CVPR2023) 2PCNet: Two-Phase Consistency Training for Day-to-Night Unsupervised Domain Adaptive Object Detection [arxiv link][code|official]

  • Harmonious-Teacher(CVPR2023) Harmonious Teacher for Cross-Domain Object Detection [paper link][code|official]

  • Scenes100(CVPR2023) Object Detection With Self-Supervised Scene Adaptation [paper link][code|official]

  • CIGAR(CVPR2023) CIGAR: Cross-Modality Graph Reasoning for Domain Adaptive Object Detection [paper link][code is not available]

  • 👍BiADT(ICCV2023) Bidirectional Alignment for Domain Adaptive Detection with Transformers [paper link][pdf link][code|official]

  • LGCL(BMVC2024)(arxiv2024.10) Improving Object Detection via Local-global Contrastive Learning [arxiv link][project link][code|official][Huawei Noah's Ark Lab + University of Edinburgh + University of Birmingham]

*********************************

③ ⭐⭐Domain Adaptation for Semantic Segmentation

  • FCNs in the Wild(arxiv2016) FCNs in the Wild: Pixel-level Adversarial and Constraint-based Adaptation [paper link][both global and category specific adaptation techniques, pioneering]

  • CDA(ICCV2017) Curriculum Domain Adaptation for Semantic Segmentation of Urban Scenes [paper link][code|official][curriculum domain adaptation]

  • CyCADA(ICML2018) CyCADA: Cycle-Consistent Adversarial Domain Adaptation [paper link][adversarial training]

  • AdaptSegNet(CVPR2018) Learning to Adapt Structured Output Space for Semantic Segmentation [paper link][code|official][adversarial learning]

  • ADVENT(CVPR2019 oral) ADVENT: Adversarial Entropy Minimization for Domain Adaptation in Semantic Segmentation [paper link][code|official][adversarial training]

  • BDL(CVPR2019) Bidirectional Learning for Domain Adaptation of Semantic Segmentation [paper link][code|official][adversarial training]

  • TGCF-DA(ICCV2019) Self-Ensembling With GAN-Based Data Augmentation for Domain Adaptation in Semantic Segmentation [paper link][GAN-Based Data Augmentation]

  • Adapt-Seg(ICCV2019 Oral) Domain Adaptation for Structured Output via Discriminative Patch Representations [paper link][project link][adversarial learning scheme]

  • FDA(CVPR2020) FDA: Fourier Domain Adaptation for Semantic Segmentation [paper link][code|official]

  • FADA(ECCV2020) Classes Matter: A Fine-Grained Adversarial Approach to Cross-Domain Semantic Segmentation [paper link][codes|official PyTorch][self-training]

  • ❤ProDA(CVPR2021) Prototypical Pseudo Label Denoising and Target Structure Learning for Domain Adaptive Semantic Segmentation [paper link][codes|official PyTorch][Use prototypes to weight pseudo-labels]

  • (CVPR2021) Coarse-To-Fine Domain Adaptive Semantic Segmentation With Photometric Alignment and Category-Center Regularization [paper link][self-training]

  • PixMatch(CVPR2021) PixMatch: Unsupervised Domain Adaptation via Pixelwise Consistency Training [paper link][codes|official PyTorch][self-training]

  • DA-SAC(CVPR2021) Self-Supervised Augmentation Consistency for Adapting Semantic Segmentation [paper link][codes|official PyTorch][self-training]

  • ❤DAFormer(CVPR2022) DAFormer: Improving Network Architectures and Training Strategies for Domain-Adaptive Semantic Segmentation [paper link][codes|official PyTorch][Rare Class Sampling (RCS) + Thing-Class ImageNet Feature Distance (FD) + Learning Rate Warmup]

  • SimT(CVPR2022) SimT: Handling Open-Set Noise for Domain Adaptive Semantic Segmentation [paper link][codes|official PyTorch][self-training]

  • CPSL(CVPR2022) Class-Balanced Pixel-Level Self-Labeling for Domain Adaptive Semantic Segmentation [paper link][codes|official PyTorch][self-training]

  • ❤ProCA(ECCV2022) Prototypical Contrast Adaptation for Domain Adaptive Semantic Segmentation [paper link][codes|official PyTorch][Prototype to feature contrastive]

  • ❤HRDA(ECCV2022) HRDA: Context-Aware High-Resolution Domain-Adaptive Semantic Segmentation [paper link][codes|official PyTorch][Based on DAFormer]

  • DecoupleNet(ECCV2022) DecoupleNet: Decoupled Network for Domain Adaptive Semantic Segmentation [paper link][codes|official PyTorch][self-training]

  • DDB(NIPS2022) Deliberated Domain Bridging for Domain Adaptive Semantic Segmentation [paper link][codes|official PyTorch][self-training]

  • BiSMAP(ACMMM2022) Bidirectional Self-Training with Multiple Anisotropic Prototypes for Domain Adaptive Semantic Segmentation [paper link][Use gaussian mixture model as prototypes to generate pseudo-labels]

  • SePiCo(TPAMI2023) SePiCo: Semantic-Guided Pixel Contrast for Domain Adaptive Semantic Segmentation [paper link][codes|official PyTorch][Contrastive with centroid, memory band and gaussian]

  • WSDA_semantic(MTA2023) On exploring weakly supervised domain adaptation strategies for semantic segmentation using synthetic data [paper link][code|official]

  • DPPASS(CVPR2023) Both Style and Distortion Matter: Dual-Path Unsupervised Domain Adaptation for Panoramic Semantic Segmentation [paper link][arxiv link][project link]

  • DIGA(CVPR2023) Dynamically Instance-Guided Adaptation: A Backward-Free Approach for Test-Time Domain Adaptive Semantic Segmentation [paper link][code|official]

*********************************

④ ⭐Domain Generalization Methods

Unlike other related learning problems such as domain adaptation or transfer learning, DG considers the scenarios where target data is inaccessible during model learning.

  • (NIPS2011) Generalizing from Several Related Classification Tasks to a New Unlabeled Sample [paper link][Universitat Potsdam + University of Michigan][the problem of domain generalization (DG) was firstly introduced.]

  • (CVPR2011) Unbiased Look at Dataset Bias [paper link][pdf link][MIT + CMU][A seminal work raised attention on the cross-domain generalization issue in computer vision. It performed a thorough investigation into the cross-dataset generalization performance of object recognition models using six popular benchmark datasets. Their findings suggested that dataset biases, which are difficult to avoid, can lead to poor generalization performance.]

  • (CVPR2012) Undoing the Damage of Dataset Bias [paper link][pdf link][MIT + CMU][It targeted the cross-dataset generalization problem in classification and detection tasks, and proposed to learn domain-specific bias vectors and domainagnostic weight vectors based on support vector machine (SVM) classifiers.]

  • (ICML2013) Domain Generalization via Invariant Feature Representation [paper link][arxiv link][MPII + ETH Zurich][the term domain generalization was later coined in this paper after (NIPS2011)]

▶4.1 Image Classification

  • PCL(CVPR2022) PCL: Proxy-Based Contrastive Learning for Domain Generalization [paper link][CUHK]

  • ITTA(CVPR2023) Improved Test-Time Adaptation for Domain Generalization [paper link][code|official]

  • DFF(Deep Frequency Filtering)(CVPR2023) Deep Frequency Filtering for Domain Generalization [paper link][arxiv link]

  • DART(Diversify-Aggregate-Repeat Training)(CVPR2023) DART: Diversify-Aggregate-Repeat Training Improves Generalization of Neural Networks [paper link][arxiv link][code|official]

  • DCG(CVPR2023) Improving Generalization With Domain Convex Game [paper link][arxiv link][code|official]

  • NICO++(CVPR2023) NICO++: Towards Better Benchmarking for Domain Generalization [paper link][arxiv link][code|official]

  • SAGM(CVPR2023) Sharpness-Aware Gradient Matching for Domain Generalization [paper link][arxiv link][code|official]

  • DAC-P & DAC-SC(CVPR2023) Decompose, Adjust, Compose: Effective Normalization by Playing With Frequency for Domain Generalization [paper link][arxiv link]

  • Pro-RandConv(CVPR2023) Progressive Random Convolutions for Single Domain Generalization [paper link][arxiv link]

  • 👍OKDPH(CVPR2023) Generalization Matters: Loss Minima Flattening via Parameter Hybridization for Efficient Online Knowledge Distillation [paper link][arxiv link][code|official][online knowledge distillation (OKD)]

  • FedDG-GA(CVPR2023) Federated Domain Generalization With Generalization Adjustment [paper link][code|official]

  • MCL(CVPR2023) Meta-Causal Learning for Single Domain Generalization [paper link][arxiv link]

  • MAD(Modality-Agnostic Debiasing)(CVPR2023) Modality-Agnostic Debiasing for Single Domain Generalization [paper link][arxiv link]

  • ALOFT(CVPR2023) ALOFT: A Lightweight MLP-Like Architecture With Dynamic Low-Frequency Transform for Domain Generalization [paper link][arxiv link][code|official]

  • DN2A(CVPR2023) Promoting Semantic Connectivity: Dual Nearest Neighbors Contrastive Learning for Unsupervised Domain Generalization [paper link]

▶4.2 Object Detection

  • Single-DGOD(CVPR2022) Single-Domain Generalized Object Detection in Urban Scene via Cyclic-Disentangled Self-Distillation [paper link][code|official][the first SDG object detection method]

  • H2FA_R-CNN(CVPR2022) H2FA R-CNN: Holistic and Hierarchical Feature Alignment for Cross-Domain Weakly Supervised Object Detection [paper link][code|official][Baidu Research, Cross-domain weakly supervised object detection (CDWSOD)]

  • DomainGen(CVPR2023) CLIP the Gap: A Single Domain Generalization Approach for Object Detection [paper link][arxiv link][code|official][By this paper, the literature on SDG object detection remains almost non-existent]

  • DG-BEV(CVPR2023) Towards Domain Generalization for Multi-View 3D Object Detection in Bird-Eye-View [paper link][arxiv link][this is the first systematic study to explore a domain generalization method for MV3D-Det]

▶4.3 Semantic Segmentation

*********************************

⑤ ⭐Source-Free Domain Adaptation Methods

▶5.1 Image Classification

  • 👍USFDA(CVPR2020) Universal Source-Free Domain Adaptation [paper link][project link]

  • (WACV2021) Domain Impression: A Source Data Free Domain Adaptation Method [paper link]

  • G-SFDA(ICCV2021) Generalized Source-Free Domain Adaptation [paper link][project link][code|official]

  • A2Net(ICCV2021) Adaptive Adversarial Network for Source-Free Domain Adaptation [paper link][code|official (can not be used)]

  • CAiDA(NIPS2021) Confident Anchor-Induced Multi-Source Free Domain Adaptation [paper link][code|official]

  • SFDA_neighbors(NIPS2021) Exploiting the Intrinsic Neighborhood Structure for Source-free Domain Adaptation [paper link][code|official]

  • ProxyMix(arxiv2022.05) ProxyMix: Proxy-based Mixup Training with Label Refinery for Source-Free Domain Adaptation [arxiv link][code|official]

  • Mixup-SFDA(ICML2022) Balancing Discriminability and Transferability for Source-Free Domain Adaptation [paper link][project link][code|official]

  • CoWA-JMDS(ICML2022) Confidence Score for Source-Free Unsupervised Domain Adaptation [paper link][code|official]

  • U-SFAN(ECCV2022) Uncertainty-Guided Source-Free Domain Adaptation [paper link][arxiv link][code|official]

  • Variational_Model_Perturbation(NIPS2022) Variational Model Perturbation for Source-Free Domain Adaptation [paper link][code|official]

  • AaD_SFDA(NIPS2022 Spotlight) Attracting and Dispersing: A Simple Approach for Source-free Domain Adaptation [openreview link][arxiv link][code|official]

  • SFDA-DE(CVPR2022) Source-Free Domain Adaptation via Distribution Estimation [paper link]

  • DIPE (Domain-Invariant Parameter Exploring)(CVPR2022) Exploring Domain-Invariant Parameters for Source Free Domain Adaptation [paper link]

  • VDB(CVPR2022) Visual Domain Bridge: A Source-Free Domain Adaptation for Cross-Domain Few-Shot Learning [paper link][code|official][adding Few-Shot Learning]

  • survey(arxiv2023.02) A Comprehensive Survey on Source-free Domain Adaptation [arxiv link]

  • C-SFDA(CVPR2023) C-SFDA: A Curriculum Learning Aided Self-Training Framework for Efficient Source Free Domain Adaptation [arxiv link][project link]

  • MHPL(CVPR2023) MHPL: Minimum Happy Points Learning for Active Source Free Domain Adaptation [paper link]

  • GPL(CVPR2023) Guiding Pseudo-Labels With Uncertainty Estimation for Source-Free Unsupervised Domain Adaptation [paper link][arxiv link][code|official]

  • CRCo(CVPR2023) Class Relationship Embedded Learning for Source-Free Unsupervised Domain Adaptation [paper link][code|official]

▶5.2 Object Detection

▶5.3 Semantic Segmentation

  • SFDA(MICCAI2020) Source-Relaxed Domain Adaptation for Image Segmentation [paper link][code|official]

  • 👍SFDA-Seg(CVPR2021) Source-Free Domain Adaptation for Semantic Segmentation [paper link][arxiv link][only a well-trained source model and an unlabeled target domain dataset are available for adaptation]

  • 👍STPL(CVPR2023) Spatio-Temporal Pixel-Level Contrastive Learning-Based Source-Free Domain Adaptation for Video Semantic Segmentation [paper link][arxiv link][code|official]

*********************************

⑥ ⭐ Semi-Supervised Domain Adaptation

  • AdaEmbed(arxiv2024.01) AdaEmbed: Semi-supervised Domain Adaptation in the Embedding Space [arxiv link][Stanford University]

*********************************

⑦ ⭐Domain Adaptation for Other Fields

  • SSDA3D(AAAI2023) SSDA3D: Semi-supervised Domain Adaptation for 3D Object Detection from Point Cloud [paper link][codes|official (not released)][Domain Adaptation for 3D Object Detection]

  • CMOM(WACV2023) Domain Adaptive Video Semantic Segmentation via Cross-Domain Moving Object Mixing [paper link][Domain Adaptation for Video Semantic Segmentation]

  • TTA-COPE(CVPR2023) TTA-COPE: Test-Time Adaptation for Category-Level Object Pose Estimation [paper link][]