ICCV 2021 accepted papers on Label-efficient and robust learning
273 of the 1,612 papers accepted at ICCV 2021 (16.9%) are on Label-efficient and robust learning, compared with 15.0% at ICCV 2019. They fall into 9 topics.
See ICCV 2021 in the interactive map
Submitting to ICCV? 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 · 83 papers
- TransReID: Transformer-Based Object Re-Identification (1,396 citations)
- Asymmetric Loss for Multi-Label Classification (867 citations)
- DenseTNT: End-to-End Trajectory Prediction From Dense Goal Sets (559 citations)
- GLoRIA: A Multimodal Global-Local Representation Learning Framework for Label-Efficient Medical Image Recognition (545 citations)
- With a Little Help From My Friends: Nearest-Neighbor Contrastive Learning of Visual Representations (444 citations)
- Counterfactual Attention Learning for Fine-Grained Visual Categorization and Re-Identification (420 citations)
- Channel Augmented Joint Learning for Visible-Infrared Recognition (413 citations)
- TrivialAugment: Tuning-Free Yet State-of-the-Art Data Augmentation (372 citations)
- Parametric Contrastive Learning (334 citations)
- Learning by Aligning: Visible-Infrared Person Re-Identification Using Cross-Modal Correspondences (325 citations)
- ICE: Inter-Instance Contrastive Encoding for Unsupervised Person Re-Identification (310 citations)
- OpenGAN: Open-Set Recognition via Open Data Generation (236 citations)
and 71 more in the interactive map.
Domain adaptation and generalization · 62 papers
- Learning To Diversify for Single Domain Generalization (314 citations)
- Generalized Source-Free Domain Adaptation (303 citations)
- SelfReg: Self-Supervised Contrastive Regularization for Domain Generalization (290 citations)
- Adaptive Adversarial Network for Source-Free Domain Adaptation (224 citations)
- A Simple Feature Augmentation for Domain Generalization (220 citations)
- OVANet: One-vs-All Network for Universal Domain Adaptation (179 citations)
- IDM: An Intermediate Domain Module for Domain Adaptive Person Re-ID (167 citations)
- Active Domain Adaptation via Clustering Uncertainty-Weighted Embeddings (153 citations)
- Domain Adaptive Semantic Segmentation With Self-Supervised Depth Estimation (149 citations)
- SENTRY: Selective Entropy Optimization via Committee Consistency for Unsupervised Domain Adaptation (126 citations)
- Generalize Then Adapt: Source-Free Domain Adaptive Semantic Segmentation (114 citations)
- Uncertainty-Aware Pseudo Label Refinery for Domain Adaptive Semantic Segmentation (113 citations)
and 50 more in the interactive map.
Self-supervised and contrastive · 36 papers
- Emerging Properties in Self-Supervised Vision Transformers (7,104 citations)
- An Empirical Study of Training Self-Supervised Vision Transformers (1,947 citations)
- Big Self-Supervised Models Advance Medical Image Classification (723 citations)
- Seasonal Contrast: Unsupervised Pre-Training From Uncurated Remote Sensing Data (397 citations)
- Geography-Aware Self-Supervised Learning (325 citations)
- Co2L: Contrastive Continual Learning (247 citations)
- SignBERT: Pre-Training of Hand-Model-Aware Representation for Sign Language Recognition (161 citations)
- On Compositions of Transformations in Contrastive Self-Supervised Learning (148 citations)
- Weakly Supervised Contrastive Learning (148 citations)
- Efficient Visual Pretraining With Contrastive Detection (145 citations)
- Self-Supervised Neural Networks for Spectral Snapshot Compressive Imaging (144 citations)
- On Feature Decorrelation in Self-Supervised Learning (140 citations)
and 24 more in the interactive map.
Zero-/few-shot and meta-learning · 26 papers
- Meta-Baseline: Exploring Simple Meta-Learning for Few-Shot Learning (643 citations)
- Relational Embedding for Few-Shot Classification (299 citations)
- FREE: Feature Refinement for Generalized Zero-Shot Learning (255 citations)
- Z-Score Normalization, Hubness, and Few-Shot Learning (226 citations)
- Meta-Learning With Task-Adaptive Loss Function for Few-Shot Learning (168 citations)
- Semantics Disentangling for Generalized Zero-Shot Learning (156 citations)
- Binocular Mutual Learning for Improving Few-Shot Classification (114 citations)
- Mixture-Based Feature Space Learning for Few-Shot Image Classification (107 citations)
- Universal Representation Learning From Multiple Domains for Few-Shot Classification (107 citations)
- Elaborative Rehearsal for Zero-Shot Action Recognition (100 citations)
- Variational Feature Disentangling for Fine-Grained Few-Shot Classification (86 citations)
- Task-Aware Part Mining Network for Few-Shot Learning (79 citations)
and 14 more in the interactive map.
Semi-supervised learning · 24 papers
- CoMatch: Semi-Supervised Learning With Contrastive Graph Regularization (313 citations)
- Semi-Supervised Semantic Segmentation With Pixel-Level Contrastive Learning From a Class-Wise Memory Bank (288 citations)
- Pixel Contrastive-Consistent Semi-Supervised Semantic Segmentation (212 citations)
- A Unified Objective for Novel Class Discovery (210 citations)
- A Simple Baseline for Semi-Supervised Semantic Segmentation With Strong Data Augmentation (158 citations)
- Re-Distributing Biased Pseudo Labels for Semi-Supervised Semantic Segmentation: A Baseline Investigation (156 citations)
- Spatial Uncertainty-Aware Semi-Supervised Crowd Counting (122 citations)
- Semi-Supervised Learning of Visual Features by Non-Parametrically Predicting View Assignments With Support Samples (115 citations)
- C3-SemiSeg: Contrastive Semi-Supervised Segmentation via Cross-Set Learning and Dynamic Class-Balancing (80 citations)
- Iterative Label Cleaning for Transductive and Semi-Supervised Few-Shot Learning (80 citations)
- Semi-Supervised Active Learning With Temporal Output Discrepancy (74 citations)
- Pseudo-Loss Confidence Metric for Semi-Supervised Few-Shot Learning (69 citations)
and 12 more in the interactive map.
Knowledge distillation · 22 papers
- Channel-Wise Knowledge Distillation for Dense Prediction (520 citations)
- Self-Knowledge Distillation With Progressive Refinement of Targets (296 citations)
- Densely Guided Knowledge Distillation Using Multiple Teacher Assistants (161 citations)
- Exploring Inter-Channel Correlation for Diversity-Preserved Knowledge Distillation (136 citations)
- Online Knowledge Distillation for Efficient Pose Estimation (124 citations)
- Student Customized Knowledge Distillation: Bridging the Gap Between Student and Teacher (90 citations)
- Compressing Visual-Linguistic Model via Knowledge Distillation (85 citations)
- Distilling Holistic Knowledge With Graph Neural Networks (72 citations)
- Adaptive Curriculum Learning (53 citations)
- Learning an Augmented RGB Representation With Cross-Modal Knowledge Distillation for Action Detection (48 citations)
- MixMix: All You Need for Data-Free Compression Are Feature and Data Mixing (43 citations)
- Distilling Global and Local Logits With Densely Connected Relations (42 citations)
and 10 more in the interactive map.
OOD and anomaly detection · 10 papers
- The Many Faces of Robustness: A Critical Analysis of Out-of-Distribution Generalization (1,472 citations)
- Divide-and-Assemble: Learning Block-Wise Memory for Unsupervised Anomaly Detection (215 citations)
- Entropy Maximization and Meta Classification for Out-of-Distribution Detection in Semantic Segmentation (170 citations)
- A Hierarchical Transformation-Discriminating Generative Model for Few Shot Anomaly Detection (111 citations)
- Semantically Coherent Out-of-Distribution Detection (97 citations)
- Learning Unsupervised Metaformer for Anomaly Detection (96 citations)
- CrossNorm and SelfNorm for Generalization Under Distribution Shifts (85 citations)
- NGC: A Unified Framework for Learning With Open-World Noisy Data (76 citations)
- Triggering Failures: Out-of-Distribution Detection by Learning From Local Adversarial Attacks in Semantic Segmentation (49 citations)
- CODEs: Chamfer Out-of-Distribution Examples Against Overconfidence Issue (33 citations)
Noisy labels · 7 papers
- Learning From Noisy Data With Robust Representation Learning (123 citations)
- Learning Fast Sample Re-Weighting Without Reward Data (106 citations)
- Learning With Noisy Labels via Sparse Regularization (85 citations)
- Me-Momentum: Extracting Hard Confident Examples From Noisily Labeled Data (32 citations)
- Searching for Robustness: Loss Learning for Noisy Classification Tasks (14 citations)
- Adaptive Label Noise Cleaning With Meta-Supervision for Deep Face Recognition (12 citations)
- Switchable K-Class Hyperplanes for Noise-Robust Representation Learning (6 citations)
Test-time adaptation · 3 papers
- Better Aggregation in Test-Time Augmentation (280 citations)
- Predicting With Confidence on Unseen Distributions (104 citations)
- Deep Matching Prior: Test-Time Optimization for Dense Correspondence (20 citations)
