ECCV 2020 accepted papers on Label-efficient and robust learning
227 of the 1,358 papers accepted at ECCV 2020 (16.7%) are on Label-efficient and robust learning, compared with 13.3% at ECCV 2018. They fall into 9 topics.
See ECCV 2020 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
Recognition and long-tail · 79 papers
- Contrastive Multiview Coding (2,060 citations)
- Dynamic Dual-Attentive Aggregation Learning for Visible-Infrared Person Re-Identification (586 citations)
- A Metric Learning Reality Check (502 citations)
- Fine-Grained Visual Classification via Progressive Multi-Granularity Training of Jigsaw Patches (433 citations)
- Unifying Deep Local and Global Features for Image Search (412 citations)
- Identity-Guided Human Semantic Parsing for Person Re-Identification (395 citations)
- Distribution-Balanced Loss for Multi-Label Classification in Long-Tailed Datasets (293 citations)
- MTI-Net: Multi-Scale Task Interaction Networks for Multi-Task Learning (281 citations)
- Feature Space Augmentation for Long-Tailed Data (274 citations)
- Learning Open Set Network with Discriminative Reciprocal Points (271 citations)
- Sub-center ArcFace: Boosting Face Recognition by Large-scale Noisy Web Faces (261 citations)
- Attention-Driven Dynamic Graph Convolutional Network for Multi-Label Image Recognition (254 citations)
and 67 more in the interactive map.
Domain adaptation and generalization · 58 papers
- Minimum Class Confusion for Versatile Domain Adaptation (464 citations)
- Learning to Generate Novel Domains for Domain Generalization (436 citations)
- Classes Matter: A Fine-grained Adversarial Approach to Cross-domain Semantic Segmentation (332 citations)
- Instance Adaptive Self-Training for Unsupervised Domain Adaptation (311 citations)
- Learning to Optimize Domain Specific Normalization for Domain Generalization (236 citations)
- Multiple Expert Brainstorming for Domain Adaptive Person Re-identification (223 citations)
- Unsupervised Domain Adaptation with Noise Resistible Mutual-Training for Person Re-identification (214 citations)
- Learning from Extrinsic and Intrinsic Supervisions for Domain Generalization (210 citations)
- Learning to Balance Specificity and Invariance for In and Out of Domain Generalization (195 citations)
- Joint Visual and Temporal Consistency for Unsupervised Domain Adaptive Person Re-Identification (174 citations)
- Dual Mixup Regularized Learning for Adversarial Domain Adaptation (171 citations)
- Learning to Detect Open Classes for Universal Domain Adaptation (169 citations)
and 46 more in the interactive map.
Zero-/few-shot and meta-learning · 24 papers
- Rethinking Few-shot Image Classification: A Good Embedding is All You Need? (1,014 citations)
- Negative Margin Matters: Understanding Margin in Few-shot Classification (397 citations)
- A Broader Study of Cross-Domain Few-Shot Learning (374 citations)
- Latent Embedding Feedback and Discriminative Features for Zero-Shot Classification (289 citations)
- Embedding Propagation: Smoother Manifold for Few-Shot Classification (233 citations)
- Region Graph Embedding Network for Zero-Shot Learning (189 citations)
- Associative Alignment for Few-shot Image Classification (168 citations)
- ProxyNCA++: Revisiting and Revitalizing Proxy Neighborhood Component Analysis (166 citations)
- An Ensemble of Epoch-wise Empirical Bayes for Few-shot Learning (149 citations)
- Leveraging Seen and Unseen Semantic Relationships for Generative Zero-Shot Learning (146 citations)
- Selecting Relevant Features from a Multi-domain Representation for Few-shot Classification (127 citations)
- Invertible Zero-Shot Recognition Flows (123 citations)
and 12 more in the interactive map.
Semi-supervised learning · 22 papers
- Guided Collaborative Training for Pixel-wise Semi-Supervised Learning (288 citations)
- Consistency-based Semi-supervised Active Learning: Towards Minimizing Labeling Cost (199 citations)
- Naive-Student: Leveraging Semi-Supervised Learning in Video Sequences for Urban Scene Segmentation (169 citations)
- Semi-Supervised Segmentation based on Error-Correcting Supervision (125 citations)
- FeatMatch: Feature-Based Augmentation for Semi-Supervised Learning (118 citations)
- Multi-Task Curriculum Framework for Open-Set Semi-Supervised Learning (110 citations)
- Semi-Supervised Crowd Counting via Self-Training on Surrogate Tasks (93 citations)
- Semi-supervised Semantic Segmentation via Strong-weak Dual-branch Network (89 citations)
- Adversarial Self-Supervised Learning for Semi-Supervised 3D Action Recognition (79 citations)
- Learning to Count in the Crowd from Limited Labeled Data (77 citations)
- Active Crowd Counting with Limited Supervision (55 citations)
- Transformation Consistency Regularization – A Semi-Supervised Paradigm for Image-to-Image Translation (33 citations)
and 10 more in the interactive map.
Knowledge distillation · 22 papers
- Knowledge Distillation Meets Self-Supervision (316 citations)
- Learning From Multiple Experts: Self-paced Knowledge Distillation for Long-tailed Classification (311 citations)
- Feature Normalized Knowledge Distillation for Image Classification (106 citations)
- Robust Re-Identification by Multiple Views Knowledge Distillation (74 citations)
- Circumventing Outliers of AutoAugment with Knowledge Distillation (69 citations)
- Online Ensemble Model Compression using Knowledge Distillation (68 citations)
- Matching Guided Distillation (66 citations)
- Knowledge Transfer via Dense Cross-Layer Mutual-Distillation (63 citations)
- Exclusivity-Consistency Regularized Knowledge Distillation for Face Recognition (62 citations)
- Optical Flow Distillation: Towards Efficient and Stable Video Style Transfer (62 citations)
- Local Correlation Consistency for Knowledge Distillation (55 citations)
- Prime-Aware Adaptive Distillation (49 citations)
and 10 more in the interactive map.
Self-supervised and contrastive · 9 papers
- SCAN: Learning to Classify Images without Labels (579 citations)
- Feature-metric Loss for Self-supervised Learning of Depth and Egomotion (260 citations)
- The Devil is in the Details: Self-Supervised Attention for Vehicle Re-Identification (173 citations)
- When Does Self-supervision Improve Few-shot Learning? (165 citations)
- GATCluster: Self-Supervised Gaussian-Attention Network for Image Clustering (85 citations)
- Cyclic Functional Mapping: Self-supervised Correspondence between Non-isometric Deformable Shapes (61 citations)
- Self-supervised Outdoor Scene Relighting (53 citations)
- Extending and Analyzing Self-Supervised Learning Across Domains (37 citations)
- Unsupervised Deep Metric Learning with Transformed Attention Consistency and Contrastive Clustering Loss (25 citations)
OOD and anomaly detection · 7 papers
- Few-Shot Scene-Adaptive Anomaly Detection (171 citations)
- Encoding Structure-Texture Relation with P-Net for Anomaly Detection in Retinal Images (154 citations)
- Backpropagated Gradient Representations for Anomaly Detection (79 citations)
- Neural Batch Sampling with Reinforcement Learning for Semi-Supervised Anomaly Detection (37 citations)
- EGDCL: An Adaptive Curriculum Learning Framework for Unbiased Glaucoma Diagnosis (24 citations)
- Multiple Class Novelty Detection Under Data Distribution Shift (15 citations)
- Utilizing Patch-level Category Activation Patterns for Multiple Class Novelty Detection (7 citations)
Noisy labels · 4 papers
- Learning Noise-Aware Encoder-Decoder from Noisy Labels by Alternating Back-Propagation for Saliency Detection (64 citations)
- NoiseRank: Unsupervised Label Noise Reduction with Dependence Models (36 citations)
- Suppressing Mislabeled Data via Grouping and Self-Attention (34 citations)
- Graph convolutional networks for learning with few clean and many noisy labels (26 citations)
