mlsanityData to 7 Oct 2026

Few-shot Open-set Recognition Using Background as Unknowns

arXiv 2207.09059 ↗First posted 19 Jul 2022cs.CVcited by 6

Not on the map. Why?

Abstract (arXiv)

Few-shot open-set recognition aims to classify both seen and novel images given only limited training data of seen classes. The challenge of this task is that the model is required not only to learn a discriminative classifier to classify the pre-defined classes with few training data but also to reject inputs from unseen classes that never appear at training time. In this paper, we propose to solve the problem from two novel aspects. First, instead of learning the decision boundaries between seen classes, as is done in standard close-set classification, we reserve space for unseen classes, such that images located in these areas are recognized as the unseen classes. Second, to effectively learn such decision boundaries, we propose to utilize the background features from seen classes. As these background regions do not significantly contribute to the decision of close-set classification, it is natural to use them as the pseudo unseen classes for classifier learning. Our extensive experiments show that our proposed method not only outperforms multiple baselines but also sets new state-of-the-art results on three popular benchmarks, namely tieredImageNet, miniImageNet, and Caltech-USCD Birds-200-2011 (CUB).

Key facts. Cited by 6 papers in the map (data through 7 Oct 2026). Builds on 41 of its 58 references that are in the map. First posted on arXiv 19 Jul 2022.

Builds on

41 of 58 references.

References · 41

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,160
  3. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,267
  4. 2017Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networkscited by 4,861
  5. 2015Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classificationcited by 3,590
  6. 2017Prototypical Networks for Few-shot Learningcited by 3,001
  7. 2015Learning Deep Features for Discriminative Localizationcited by 2,430
  8. 2016Matching Networks for One Shot Learningcited by 2,246
  9. 2017Learning to Compare: Relation Network for Few-Shot Learningcited by 1,139
  10. 2015Towards Open Set Deep Networkscited by 533
  11. 2018Meta-Learning with Latent Embedding Optimizationcited by 476
  12. 2018Meta-Learning for Semi-Supervised Few-Shot Classificationcited by 456
29 more
  1. 2018Meta-Transfer Learning for Few-Shot Learningcited by 306
  2. 2018Few-Shot Learning via Embedding Adaptation with Set-to-Set Functionscited by 293
  3. 2019Cross Attention Network for Few-shot Classificationcited by 218
  4. 2019CANet: Class-Agnostic Segmentation Networks with Iterative Refinement and Attentive Few-Shot Learningcited by 217
  5. 2021Few-Shot Incremental Learning with Continually Evolved Classifierscited by 184
  6. 2017Generative OpenMax for Multi-Class Open Set Classificationcited by 157
  7. 2018Classification-Reconstruction Learning for Open-Set Recognitioncited by 146
  8. 2019C2AE: Class Conditioned Auto-Encoder for Open-set Recognitioncited by 142
  9. 2019Charting the Right Manifold: Manifold Mixup for Few-shot Learningcited by 135
  10. 2021Learning Placeholders for Open-Set Recognitioncited by 121
  11. 2020Conditional Gaussian Distribution Learning for Open Set Recognitioncited by 104
  12. 2019Finding Task-Relevant Features for Few-Shot Learning by Category Traversalcited by 101
  13. 2020CRNet: Cross-Reference Networks for Few-Shot Segmentationcited by 101
  14. 2020Convolutional Networks with Dense Connectivitycited by 98
  15. 2021Contrastive Embedding for Generalized Zero-Shot Learningcited by 94
  16. 2020TaskNorm: Rethinking Batch Normalization for Meta-Learningcited by 50
  17. 2020Few-Shot Open-Set Recognition using Meta-Learningcited by 50
  18. 2020DeepEMD: Differentiable Earth Mover's Distance for Few-Shot Learningcited by 49
  19. 2021Few-shot Segmentation with Optimal Transport Matching and Message Flowcited by 36
  20. 2019Learning Adaptive Classifiers Synthesis for Generalized Few-Shot Learningcited by 29
  21. 2021Few-shot Open-set Recognition by Transformation Consistencycited by 29
  22. 2021CycleSegNet: Object Co-segmentation with Cycle Refinement and Region Correspondencecited by 19
  23. 2021Meta Navigator: Search for a Good Adaptation Policy for Few-shot Learningcited by 19
  24. 2022EASY: Ensemble Augmented-Shot Y-shaped Learning: State-Of-The-Art Few-Shot Classification with Simple Ingredientscited by 15
  25. 2020Compositional Prototype Network with Multi-view Comparision for Few-Shot Point Cloud Semantic Segmentationcited by 12
  26. 2021M2IOSR: Maximal Mutual Information Open Set Recognitioncited by 12
  27. 2022Efficient Few-Shot Object Detection via Knowledge Inheritancecited by 12
  28. 2020Open Set Recognition with Conditional Probabilistic Generative Modelscited by 10
  29. 2017Efficient Eye Typing with 9-direction Gaze Estimationcited by 8

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 1; 2024: 3; 2025: 1; 2026: 1.

6 papers.

  1. 2024Learning Unknowns from Unknowns: Diversified Negative Prototypes Generator for Few-Shot Open-Set Recognitioncited by 6
  2. 2023Manipulating the Label Space for In-Context Classificationcited by 5
  3. 2024Towards Few-Shot Learning in the Open World: A Review and Beyondcited by 5
  4. 2025Few-Shot Generalized Category Discovery With Retrieval-Guided Decision Boundary Enhancementcited by 1
  5. 2024Few-shot Open Relation Extraction with Gaussian Prototype and Adaptive Margincited by 0
  6. 2026A Baseline Study and Benchmark for Few-Shot Open-Set Action Recognition with Feature Residual Discriminationcited by 0