WebVision Database: Visual Learning and Understanding from Web Data

Abstract

In this paper, we present a study on learning visual recognition models from large scale noisy web data. We build a new database called WebVision, which contains more than $2.4$ million web images crawled from the Internet by using queries generated from the 1,000 semantic concepts of the benchmark ILSVRC 2012 dataset. Meta information along with those web images (e.g., title, description, tags, etc.) are also crawled. A validation set and test set containing human annotated images are also provided to facilitate algorithmic development. Based on our new database, we obtain a few interesting observations: 1) the noisy web images are sufficient for training a good deep CNN model for visual recognition; 2) the model learnt from our WebVision database exhibits comparable or even better generalization ability than the one trained from the ILSVRC 2012 dataset when being transferred to new datasets and tasks; 3) a domain adaptation issue (a.k.a., dataset bias) is observed, which means the dataset can be used as the largest benchmark dataset for visual domain adaptation. Our new WebVision database and relevant studies in this work would benefit the advance of learning state-of-the-art visual models with minimum supervision based on web data.

References in corpus (6)

• Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
• Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks
• UCF101: A Dataset of 101 Human Actions Classes From Videos in The Wild
• Unsupervised Visual Representation Learning by Context Prediction
• Learning to See by Moving
• Webly Supervised Learning of Convolutional Networks

Cited by in corpus (85)

• DivideMix: Learning with Noisy Labels as Semi-supervised Learning
• Image Classification with Deep Learning in the Presence of Noisy Labels: A Survey
• Early-Learning Regularization Prevents Memorization of Noisy Labels
• Multimodal Co-learning: Challenges, Applications with Datasets, Recent Advances and Future Directions
• Self-supervised Visual Feature Learning with Deep Neural Networks: A Survey
• Masking: A New Perspective of Noisy Supervision
• Transferable Clean-Label Poisoning Attacks on Deep Neural Nets
• Hard Sample Aware Noise Robust Learning for Histopathology Image Classification
• LongReMix: Robust Learning with High Confidence Samples in a Noisy Label Environment
• Pervasive Label Errors in Test Sets Destabilize Machine Learning Benchmarks
• Integrating Expert Knowledge with Domain Adaptation for Unsupervised Fault Diagnosis
• Contrast to Divide: Self-Supervised Pre-Training for Learning with Noisy Labels
• Efficient Self-supervised Vision Transformers for Representation Learning
• Tencent ML-Images: A Large-Scale Multi-Label Image Database for Visual Representation Learning
• Learning with Noisy Labels Revisited: A Study Using Real-World Human Annotations
• Identifying Mislabeled Data using the Area Under the Margin Ranking
• BadLabel: A Robust Perspective on Evaluating and Enhancing Label-noise Learning
• CleanNet: Transfer Learning for Scalable Image Classifier Training with Label Noise
• MoPro: Webly Supervised Learning with Momentum Prototypes
• Trustable Co-label Learning from Multiple Noisy Annotators
• Provably End-to-end Label-Noise Learning without Anchor Points
• Robust Long-Tailed Learning under Label Noise
• Generalized Jensen-Shannon Divergence Loss for Learning with Noisy Labels
• Heteroskedastic and Imbalanced Deep Learning with Adaptive Regularization
• Dual Student: Breaking the Limits of the Teacher in Semi-supervised Learning
• CurriculumNet: Weakly Supervised Learning from Large-Scale Web Images
• MetaLabelNet: Learning to Generate Soft-Labels from Noisy-Labels
• Improved Person Detection on Omnidirectional Images with Non-maxima Suppression
• Dirichlet-Based Prediction Calibration for Learning with Noisy Labels
• Learning From Long-Tailed Data With Noisy Labels
• Label-PEnet: Sequential Label Propagation and Enhancement Networks for Weakly Supervised Instance Segmentation
• SSR: An Efficient and Robust Framework for Learning with Unknown Label Noise
• Towards Understanding Deep Learning from Noisy Labels with Small-Loss Criterion
• A Simple Probabilistic Method for Deep Classification under Input-Dependent Label Noise
• Webly Supervised Image Classification with Metadata: Automatic Noisy Label Correction via Visual-Semantic Graph
• A Semi-Supervised Two-Stage Approach to Learning from Noisy Labels
• Weak-shot Fine-grained Classification via Similarity Transfer
• Robust Training with Ensemble Consensus
• Coresets for Robust Training of Neural Networks against Noisy Labels
• When Do Curricula Work?
• Faster Meta Update Strategy for Noise-Robust Deep Learning
• Confidence Adaptive Regularization for Deep Learning with Noisy Labels
• Boosting Co-teaching with Compression Regularization for Label Noise
• Instance Correction for Learning with Open-set Noisy Labels
• Cap2Det: Learning to Amplify Weak Caption Supervision for Object Detection
• Weakly Supervised Learning with Side Information for Noisy Labeled Images
• Strength from Weakness: Fast Learning Using Weak Supervision
• A Unified Plug-and-Play Framework for Effective Data Denoising and Robust Abstention
• Robust and On-the-fly Dataset Denoising for Image Classification
• Bayesian Statistics Guided Label Refurbishment Mechanism: Mitigating Label Noise in Medical Image Classification
• Distant Supervision and Noisy Label Learning for Low Resource Named Entity Recognition: A Study on Hausa and Yorùbá
• DeepSinger: Singing Voice Synthesis with Data Mined From the Web
• The iMaterialist Fashion Attribute Dataset
• An Ensemble Noise-Robust K-fold Cross-Validation Selection Method for Noisy Labels
• Deep Learning Classification With Noisy Labels
• Learning to Combat Noisy Labels via Classification Margins
• Multi-label Iterated Learning for Image Classification with Label Ambiguity
• Increasing Adversarial Uncertainty to Scale Private Similarity Testing
• Safeguarded Dynamic Label Regression for Generalized Noisy Supervision
• Learning from Large-scale Noisy Web Data with Ubiquitous Reweighting for Image Classification
• Contemplating Visual Emotions: Understanding and Overcoming Dataset Bias
• A Simple yet Effective Baseline for Robust Deep Learning with Noisy Labels
• Image recognition from raw labels collected without annotators
• Co-matching: Combating Noisy Labels by Augmentation Anchoring
• NGC: A Unified Framework for Learning with Open-World Noisy Data
• Unsupervised Multi-label Dataset Generation from Web Data
• Product Image Recognition with Guidance Learning and Noisy Supervision
• Consistency Regularization Can Improve Robustness to Label Noise
• Selective Forgetting of Deep Networks at a Finer Level than Samples
• Multiplicative Reweighting for Robust Neural Network Optimization
• Unsupervised Open Domain Recognition by Semantic Discrepancy Minimization
• Webly Supervised Image Classification with Self-Contained Confidence
• Reliable Label Bootstrapping for Semi-Supervised Learning
• Sample Prior Guided Robust Model Learning to Suppress Noisy Labels
• CoDiM: Learning with Noisy Labels via Contrastive Semi-Supervised Learning
• Rethinking Noisy Label Models: Labeler-Dependent Noise with Adversarial Awareness
• A Survey on Deep Learning with Noisy Labels: How to train your model when you cannot trust on the annotations?
• Suppressing Mislabeled Data via Grouping and Self-Attention
• Knowledge Distillation with Noisy Labels for Natural Language Understanding
• Robust Temporal Ensembling for Learning with Noisy Labels
• Adaptive Hierarchical Similarity Metric Learning with Noisy Labels
• A robust approach for deep neural networks in presence of label noise: relabelling and filtering instances during training
• Friends and Foes in Learning from Noisy Labels
• A Theoretical Analysis of Learning with Noisily Labeled Data
• Learning from Web Data with Self-Organizing Memory Module