CVPR 2022 accepted papers on Label-efficient and robust learning
337 of the 2,077 papers accepted at CVPR 2022 (16.2%) are on Label-efficient and robust learning, compared with 16.1% at CVPR 2021. They fall into 9 topics.
See CVPR 2022 in the interactive map
Submitting to CVPR? Checklist →
Working on something in this area? Describe your idea in scime atlas to see the closest papers by meaning and how crowded the spot has become.
Topics and papers
Recognition and long-tail · 112 papers
- Revisiting Skeleton-Based Action Recognition (825 citations)
- AdaFace: Quality Adaptive Margin for Face Recognition (577 citations)
- DTFD-MIL: Double-Tier Feature Distillation Multiple Instance Learning for Histopathology Whole Slide Image Classification (474 citations)
- InfoGCN: Representation Learning for Human Skeleton-Based Action Recognition (442 citations)
- Transforming Model Prediction for Tracking (435 citations)
- Part-Based Pseudo Label Refinement for Unsupervised Person Re-Identification (296 citations)
- FMCNet: Feature-Level Modality Compensation for Visible-Infrared Person Re-Identification (282 citations)
- Exact Feature Distribution Matching for Arbitrary Style Transfer and Domain Generalization (241 citations)
- Generalized Category Discovery (231 citations)
- Balanced Contrastive Learning for Long-Tailed Visual Recognition (221 citations)
- Domain Generalization via Shuffled Style Assembly for Face Anti-Spoofing (213 citations)
- Learning Memory-Augmented Unidirectional Metrics for Cross-Modality Person Re-Identification (213 citations)
and 100 more in the interactive map.
Domain adaptation and generalization · 65 papers
- DAFormer: Improving Network Architectures and Training Strategies for Domain-Adaptive Semantic Segmentation (616 citations)
- Detecting Deepfakes With Self-Blended Images (455 citations)
- Cross-Domain Adaptive Teacher for Object Detection (250 citations)
- Causality Inspired Representation Learning for Domain Generalization (223 citations)
- Reusing the Task-Specific Classifier as a Discriminator: Discriminator-Free Adversarial Domain Adaptation (218 citations)
- SHIFT: A Synthetic Driving Dataset for Continuous Multi-Task Domain Adaptation (192 citations)
- Semantic-Aware Domain Generalized Segmentation (171 citations)
- Source-Free Domain Adaptation via Distribution Estimation (156 citations)
- Category Contrast for Unsupervised Domain Adaptation in Visual Tasks (151 citations)
- The Norm Must Go On: Dynamic Unsupervised Domain Adaptation by Normalization (134 citations)
- PCL: Proxy-Based Contrastive Learning for Domain Generalization (133 citations)
- Unsupervised Domain Adaptation for Nighttime Aerial Tracking (124 citations)
and 53 more in the interactive map.
Self-supervised and contrastive · 46 papers
- Self-Supervised Pre-Training of Swin Transformers for 3D Medical Image Analysis (834 citations)
- Self-Supervised Learning of Adversarial Example: Towards Good Generalizations for Deepfake Detection (293 citations)
- Blind2Unblind: Self-Supervised Image Denoising With Visible Blind Spots (193 citations)
- AP-BSN: Self-Supervised Denoising for Real-World Images via Asymmetric PD and Blind-Spot Network (163 citations)
- CVF-SID: Cyclic Multi-Variate Function for Self-Supervised Image Denoising by Disentangling Noise From Image (108 citations)
- Multi-Frame Self-Supervised Depth With Transformers (101 citations)
- When Does Contrastive Visual Representation Learning Work? (99 citations)
- DiRA: Discriminative, Restorative, and Adversarial Learning for Self-Supervised Medical Image Analysis (94 citations)
- A Self-Supervised Descriptor for Image Copy Detection (90 citations)
- Revisiting Weakly Supervised Pre-Training of Visual Perception Models (90 citations)
- Use All the Labels: A Hierarchical Multi-Label Contrastive Learning Framework (86 citations)
- Unsupervised Deraining: Where Contrastive Learning Meets Self-Similarity (85 citations)
and 34 more in the interactive map.
Semi-supervised learning · 34 papers
- Semi-Supervised Semantic Segmentation Using Unreliable Pseudo-Labels (500 citations)
- Perturbed and Strict Mean Teachers for Semi-Supervised Semantic Segmentation (314 citations)
- SimMatch: Semi-Supervised Learning With Similarity Matching (249 citations)
- ACPL: Anti-Curriculum Pseudo-Labelling for Semi-Supervised Medical Image Classification (131 citations)
- BoostMIS: Boosting Medical Image Semi-Supervised Learning With Adaptive Pseudo Labeling and Informative Active Annotation (130 citations)
- Class-Aware Contrastive Semi-Supervised Learning (119 citations)
- Semi-Supervised Semantic Segmentation With Error Localization Network (115 citations)
- Towards Semi-Supervised Deep Facial Expression Recognition With an Adaptive Confidence Margin (105 citations)
- Debiased Learning From Naturally Imbalanced Pseudo-Labels (98 citations)
- DASO: Distribution-Aware Semantics-Oriented Pseudo-Label for Imbalanced Semi-Supervised Learning (90 citations)
- Cross-Patch Dense Contrastive Learning for Semi-Supervised Segmentation of Cellular Nuclei in Histopathologic Images (89 citations)
- Cross-Model Pseudo-Labeling for Semi-Supervised Action Recognition (63 citations)
and 22 more in the interactive map.
Zero-/few-shot and meta-learning · 29 papers
- Robust Fine-Tuning of Zero-Shot Models (400 citations)
- Learning What Not To Segment: A New Perspective on Few-Shot Segmentation (330 citations)
- Joint Distribution Matters: Deep Brownian Distance Covariance for Few-Shot Classification (298 citations)
- MSDN: Mutually Semantic Distillation Network for Zero-Shot Learning (154 citations)
- Matching Feature Sets for Few-Shot Image Classification (132 citations)
- Cross-Domain Few-Shot Learning With Task-Specific Adapters (116 citations)
- Learning To Affiliate: Mutual Centralized Learning for Few-Shot Classification (107 citations)
- Generating Representative Samples for Few-Shot Classification (104 citations)
- Siamese Contrastive Embedding Network for Compositional Zero-Shot Learning (98 citations)
- BatchFormer: Learning To Explore Sample Relationships for Robust Representation Learning (95 citations)
- Task Discrepancy Maximization for Fine-Grained Few-Shot Classification (93 citations)
- IntraQ: Learning Synthetic Images With Intra-Class Heterogeneity for Zero-Shot Network Quantization (82 citations)
and 17 more in the interactive map.
Noisy labels · 17 papers
- Learning With Twin Noisy Labels for Visible-Infrared Person Re-Identification (277 citations)
- Selective-Supervised Contrastive Learning With Noisy Labels (217 citations)
- UniCon: Combating Label Noise Through Uniform Selection and Contrastive Learning (177 citations)
- The Devil Is in the Labels: Noisy Label Correction for Robust Scene Graph Generation (113 citations)
- Learning With Neighbor Consistency for Noisy Labels (92 citations)
- PNP: Robust Learning From Noisy Labels by Probabilistic Noise Prediction (86 citations)
- Large Loss Matters in Weakly Supervised Multi-Label Classification (78 citations)
- Large-Scale Pre-Training for Person Re-Identification With Noisy Labels (78 citations)
- On Learning Contrastive Representations for Learning With Noisy Labels (76 citations)
- Robust Contrastive Learning Against Noisy Views (72 citations)
- Instance-Dependent Label-Noise Learning With Manifold-Regularized Transition Matrix Estimation (69 citations)
- Robust Equivariant Imaging: A Fully Unsupervised Framework for Learning To Image From Noisy and Partial Measurements (56 citations)
and 5 more in the interactive map.
Knowledge distillation · 14 papers
- Decoupled Knowledge Distillation (960 citations)
- Anomaly Detection via Reverse Distillation From One-Class Embedding (859 citations)
- Focal and Global Knowledge Distillation for Detectors (377 citations)
- Knowledge Distillation: A Good Teacher Is Patient and Consistent (241 citations)
- Learning Multiple Adverse Weather Removal via Two-Stage Knowledge Learning and Multi-Contrastive Regularization: Toward a Unified Model (240 citations)
- Knowledge Distillation With the Reused Teacher Classifier (231 citations)
- Localization Distillation for Dense Object Detection (204 citations)
- Knowledge Distillation via the Target-Aware Transformer (138 citations)
- Self-Distillation From the Last Mini-Batch for Consistency Regularization (100 citations)
- How Many Observations Are Enough? Knowledge Distillation for Trajectory Forecasting (72 citations)
- Multi-Objective Diverse Human Motion Prediction With Knowledge Distillation (43 citations)
- Evaluation-Oriented Knowledge Distillation for Deep Face Recognition (42 citations)
and 2 more in the interactive map.
OOD and anomaly detection · 11 papers
- Self-Supervised Predictive Convolutional Attentive Block for Anomaly Detection (288 citations)
- ViM: Out-of-Distribution With Virtual-Logit Matching (273 citations)
- Catching Both Gray and Black Swans: Open-Set Supervised Anomaly Detection (185 citations)
- OoD-Bench: Quantifying and Understanding Two Dimensions of Out-of-Distribution Generalization (80 citations)
- Rethinking Reconstruction Autoencoder-Based Out-of-Distribution Detection (77 citations)
- Estimating Example Difficulty Using Variance of Gradients (59 citations)
- Neural Mean Discrepancy for Efficient Out-of-Distribution Detection (43 citations)
- Out-of-Distribution Generalization With Causal Invariant Transformations (40 citations)
- Evading the Simplicity Bias: Training a Diverse Set of Models Discovers Solutions With Superior OOD Generalization (34 citations)
- DeepFace-EMD: Re-Ranking Using Patch-Wise Earth Mover's Distance Improves Out-of-Distribution Face Identification (24 citations)
- Deep Hybrid Models for Out-of-Distribution Detection (16 citations)
Test-time adaptation · 9 papers
- Contrastive Test-Time Adaptation (262 citations)
- Parameter-Free Online Test-Time Adaptation (135 citations)
- MM-TTA: Multi-Modal Test-Time Adaptation for 3D Semantic Segmentation (61 citations)
- Sketch3T: Test-Time Training for Zero-Shot SBIR (61 citations)
- Towards Multi-Domain Single Image Dehazing via Test-Time Training (56 citations)
- Ev-TTA: Test-Time Adaptation for Event-Based Object Recognition (31 citations)
- RIO: Rotation-Equivariance Supervised Learning of Robust Inertial Odometry (29 citations)
- Fisher Information Guidance for Learned Time-of-Flight Imaging (8 citations)
- PyMiceTracking: An Open-Source Toolbox for Real-Time Behavioral Neuroscience Experiments (3 citations)
