ECCV 2022 accepted papers on Label-efficient and robust learning
268 of the 1,645 papers accepted at ECCV 2022 (16.3%) are on Label-efficient and robust learning, compared with 16.7% at ECCV 2020. They fall into 9 topics.
See ECCV 2022 in the interactive map
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
Domain adaptation and generalization · 64 papers
- HRDA: Context-Aware High-Resolution Domain-Adaptive Semantic Segmentation (328 citations)
- Domain Generalization by Mutual-Information Regularization with Pre-trained Models (153 citations)
- Prototypical Contrast Adaptation for Domain Adaptive Semantic Segmentation (147 citations)
- Style-Hallucinated Dual Consistency Learning for Domain Generalized Semantic Segmentation (115 citations)
- Adaptive Transformers for Robust Few-Shot Cross-Domain Face Anti-Spoofing (108 citations)
- Uncertainty-Guided Source-Free Domain Adaptation (83 citations)
- A Broad Study of Pre-training for Domain Generalization and Adaptation (78 citations)
- Generative Domain Adaptation for Face Anti-Spoofing (78 citations)
- BMD: A General Class-Balanced Multicentric Dynamic Prototype Strategy for Source-Free Domain Adaptation (73 citations)
- CoSMix: Compositional Semantic Mix for Domain Adaptation in 3D LiDAR Segmentation (67 citations)
- Mimic Embedding via Adaptive Aggregation: Learning Generalizable Person Re-identification (61 citations)
- Adaptive Cross-Domain Learning for Generalizable Person Re-identification (60 citations)
and 52 more in the interactive map.
Recognition and long-tail · 64 papers
- ST-P3: End-to-End Vision-Based Autonomous Driving via Spatial-Temporal Feature Learning (320 citations)
- Cross-Modality Transformer for Visible-Infrared Person Re-identification (129 citations)
- PCW-Net: Pyramid Combination and Warping Cost Volume for Stereo Matching (122 citations)
- Interpretable Image Classification with Differentiable Prototypes Assignment (119 citations)
- Optimal Transport for Label-Efficient Visible-Infrared Person Re-identification (92 citations)
- AutoMix: Unveiling the Power of Mixup for Stronger Classifiers (89 citations)
- Modality Synergy Complement Learning with Cascaded Aggregation for Visible-Infrared Person Re-identification (83 citations)
- When Counting Meets HMER: Counting-Aware Network for Handwritten Mathematical Expression Recognition (70 citations)
- VL-LTR: Learning Class-Wise Visual-Linguistic Representation for Long-Tailed Visual Recognition (67 citations)
- Differentiable Zooming for Multiple Instance Learning on Whole-Slide Images (65 citations)
- Shap-CAM: Visual Explanations for Convolutional Neural Networks Based on Shapley Value (64 citations)
- Counterfactual Intervention Feature Transfer for Visible-Infrared Person Re-identification (59 citations)
and 52 more in the interactive map.
Self-supervised and contrastive · 42 papers
- SLIP: Self-Supervision Meets Language-Image Pre-training (392 citations)
- Masked Siamese Networks for Label-Efficient Learning (306 citations)
- Decoupled Contrastive Learning (235 citations)
- Fusion from Decomposition: A Self-Supervised Decomposition Approach for Image Fusion (221 citations)
- PASS: Part-Aware Self-Supervised Pre-training for Person Re-identification (97 citations)
- UniMiSS: Universal Medical Self-Supervised Learning via Breaking Dimensionality Barrier (94 citations)
- SdAE: Self-Distillated Masked Autoencoder (70 citations)
- Differentiable Raycasting for Self-Supervised Occupancy Forecasting (68 citations)
- Image Coding for Machines with Omnipotent Feature Learning (63 citations)
- Contrastive Positive Mining for Unsupervised 3D Action Representation Learning (57 citations)
- DisCo: Remedying Self-Supervised Learning on Lightweight Models with Distilled Contrastive Learning (56 citations)
- The Challenges of Continuous Self-Supervised Learning (48 citations)
and 30 more in the interactive map.
Knowledge distillation · 27 papers
- Masked Generative Distillation (344 citations)
- Factorizing Knowledge in Neural Networks (120 citations)
- MixSKD: Self-Knowledge Distillation from Mixup for Image Recognition (94 citations)
- Self-Regulated Feature Learning via Teacher-Free Feature Distillation (66 citations)
- Prune Your Model before Distill It (58 citations)
- A Fast Knowledge Distillation Framework for Visual Recognition (56 citations)
- Switchable Online Knowledge Distillation (56 citations)
- Teaching Where to Look: Attention Similarity Knowledge Distillation for Low Resolution Face Recognition (52 citations)
- Knowledge Condensation Distillation (46 citations)
- Prediction-Guided Distillation for Dense Object Detection (42 citations)
- KD-MVS: Knowledge Distillation Based Self-Supervised Learning for Multi-View Stereo (40 citations)
- Monitored Distillation for Positive Congruent Depth Completion (38 citations)
and 15 more in the interactive map.
Semi-supervised learning · 23 papers
- ConMatch: Semi-Supervised Learning with Confidence-Guided Consistency Regularization (64 citations)
- Semi-Supervised Keypoint Detector and Descriptor for Retinal Image Matching (57 citations)
- Towards Realistic Semi-Supervised Learning (54 citations)
- OpenLDN: Learning to Discover Novel Classes for Open-World Semi-Supervised Learning (49 citations)
- Semi-Supervised Vision Transformers (42 citations)
- Unsupervised Selective Labeling for More Effective Semi-Supervised Learning (36 citations)
- RDA: Reciprocal Distribution Alignment for Robust Semi-Supervised Learning (17 citations)
- When Active Learning Meets Implicit Semantic Data Augmentation (17 citations)
- Semi-Leak: Membership Inference Attacks against Semi-Supervised Learning (16 citations)
- Optical Flow Training under Limited Label Budget via Active Learning (15 citations)
- RVSL: Robust Vehicle Similarity Learning in Real Hazy Scenes Based on Semi-Supervised Learning (14 citations)
- Unsupervised and Semi-Supervised Bias Benchmarking in Face Recognition (11 citations)
and 11 more in the interactive map.
Zero-/few-shot and meta-learning · 18 papers
- Learning Invariant Visual Representations for Compositional Zero-Shot Learning (45 citations)
- "Contributions of Shape, Texture, and Color in Visual Recognition" (42 citations)
- Adaptive Fine-Grained Sketch-Based Image Retrieval (34 citations)
- Unsupervised Few-Shot Image Classification by Learning Features into Clustering Space (30 citations)
- Rethinking Zero-Shot Action Recognition: Learning from Latent Atomic Actions (23 citations)
- Coarse-to-Fine Incremental Few-Shot Learning (17 citations)
- Fine-Grained Data Distribution Alignment for Post-Training Quantization (15 citations)
- Zero-Shot Attribute Attacks on Fine-Grained Recognition Models (4 citations)
- Adversarial Feature Augmentation for Cross-Domain Few-Shot Classification
- CLASTER: Clustering with Reinforcement Learning for Zero-Shot Action Recognition
- Contrastive Prototypical Network with Wasserstein Confidence Penalty
- Exploring Hierarchical Graph Representation for Large-Scale Zero-Shot Image Classification
and 6 more in the interactive map.
OOD and anomaly detection · 17 papers
- SPot-the-Difference Self-Supervised Pre-training for Anomaly Detection and Segmentation (752 citations)
- Natural Synthetic Anomalies for Self-Supervised Anomaly Detection and Localization (261 citations)
- Registration Based Few-Shot Anomaly Detection (230 citations)
- DSR – A Dual Subspace Re-Projection Network for Surface Anomaly Detection (201 citations)
- DICE: Leveraging Sparsification for Out-of-Distribution Detection (182 citations)
- DenseHybrid: Hybrid Anomaly Detection for Dense Open-Set Recognition (71 citations)
- Dynamic Local Aggregation Network with Adaptive Clusterer for Anomaly Detection (51 citations)
- OOD-CV: A Benchmark for Robustness to Out-of-Distribution Shifts of Individual Nuisances in Natural Images (47 citations)
- Out-of-Distribution Detection with Semantic Mismatch under Masking (32 citations)
- Predicting Is Not Understanding: Recognizing and Addressing Underspecification in Machine Learning (23 citations)
- Class Is Invariant to Context and Vice Versa: On Learning Invariance for Out-of-Distribution Generalization (15 citations)
- Hierarchical Semi-Supervised Contrastive Learning for Contamination-Resistant Anomaly Detection (11 citations)
and 5 more in the interactive map.
Noisy labels · 9 papers
- Learn from All: Erasing Attention Consistency for Noisy Label Facial Expression Recognition (283 citations)
- Neighborhood Collective Estimation for Noisy Label Identification and Correction (57 citations)
- Self-Filtering: A Noise-Aware Sample Selection for Label Noise with Confidence Penalization (57 citations)
- Teaching with Soft Label Smoothing for Mitigating Noisy Labels in Facial Expressions (44 citations)
- Centrality and Consistency: Two-Stage Clean Samples Identification for Learning with Instance-Dependent Noisy Labels (37 citations)
- Learning Semantic Correspondence with Sparse Annotations (29 citations)
- Learning with Noisy Labels by Efficient Transition Matrix Estimation to Combat Label Miscorrection (22 citations)
- Learning from Multiple Annotator Noisy Labels via Sample-Wise Label Fusion (10 citations)
- Active Label Correction Using Robust Parameter Update and Entropy Propagation
Test-time adaptation · 4 papers
- Improving Test-Time Adaptation via Shift-Agnostic Weight Regularization and Nearest Source Prototypes (62 citations)
- Context-Aware Streaming Perception in Dynamic Environments (10 citations)
- Learning Instance-Specific Adaptation for Cross-Domain Segmentation (7 citations)
- PointFix: Learning to Fix Domain Bias for Robust Online Stereo Adaptation (5 citations)
