Recent Advances in Open Set Recognition: A Survey
By Chuanxing Geng, Sheng-Jun Huang, Songcan Chen. 2018
1,029 citations recorded by the source corpus.
In real-world recognition/classification tasks, limited by various objective factors, it is usually difficult to collect training samples to exhaust all classes when training a recognizer or classifier. A more realistic scenario is open set recognition (OSR), where incomplete knowledge of the world exists at training time, and unknown classes can be submitted to an algorithm during testing, requiring the classifiers to not only accurately classify the seen classes, but also effectively deal with unseen ones. This paper provides a comprehensive survey of existing open set recognition techniques covering various aspects ranging from related definitions, representations of models, datasets, evaluation criteria, and algorithm comparisons. Furthermore, we briefly analyze the relationships between OSR and its related tasks including zero-shot, one-shot (few-shot) recognition/learning techniques, classification with reject option, and so forth. Additionally, we also review the open world recognition which can be seen as a natural extension of OSR. Importantly, we highlight the limitations of existing approaches and point out some promising subsequent research directions in this field.
Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Anomaly & Outlier Detection
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
In Computer Vision and Pattern Recognition
- Luc Van Gool
- Ming-Hsuan Yang
- Chen Change Loy
- Dacheng Tao
- Xiaogang Wang
- Bernt Schiele
- Trevor Darrell
- Yu Qiao
In Anomaly & Outlier Detection
Citations in the corpus
- Language-driven Semantic Segmentation (2022, background): “A variety of solutions have been proposed (Zhang et al., 2020b; Liu et al., 2020a; Perera et al., 2020; Zhou et al., 2021) for open-set recognition (Scheirer et al., 2012; Geng et al., 2020).”
- A Theoretical Study on Solving Continual Learning (2022, neutral): “[61] Chuanxing Geng, Sheng-jun Huang, and Songcan Chen.”
- Novel Class Discovery without Forgetting (2022, background): “Methods that perform out-of-distribution detection [32,38] and open-set learning [46,18] identify instances significantly different from the training data distribution as novel samples, but do not identify sub-groups within these identified instances automatically.”
- Out-of-Domain Detection for Natural Language Understanding in Dialog Systems (2019, neutral): “many contexts with different alias, such as “anomaly detection” [15], “one-class classification” [18], [19], “open-set recognition” [20], or “novelty detection” [21].”
- Learning Placeholders for Open-Set Recognition (2021, methodology): “There are two lines of work for open-set recognition, i.e ., discriminative models and generative models [7].”
- Deep learning and computer vision will transform entomology (2020, methodology): “However, the current methods are typically applied to relatively small datasets and do not scale well with the number of classes (96).”
- Generalized Out-of-Distribution Detection: A Survey (2021, neutral): “years [20], [21], [22], [23], [29], this survey highlights the overview of OOD detection methods and thus complements existing surveys.”
- OpenGAN: Open-Set Recognition via Open Data Generation (2021, neutral): “There are multiple lines of work addressing open-set discrimination, such as anomaly detection [9, 29, 59], outlier detection [44, 39], and open-set recognition [45, 19].”
- Towards a robust experimental framework and benchmark for lifelong language learning (2021, background): “Continual open-set classification and open-world recognition have a flavor of lifelong 93 learning [17], but they are essentially classification tasks.”
- Hybrid Models for Open Set Recognition (2020, background): “Open set classification performance can be characterized by F-score or AUROC (Area Under the ROC Curve) [7].”
- Human Action Recognition and Prediction: A Survey (2018, background): “Humanaction recognition in realworld is essentially an open set problem, which requires the model to simultaneously recognize the known action classes and reject the unknown actions (Geng et al., 2020; Bao et al., 2021).”
- Learning to Prompt for Open-Vocabulary Object Detection with Vision-Language Model (2022, background): “To alleviate the reliability of DNNs on big data and elaborate annotations, different paradigms such as semisupervised learning [34], few-shot learning [25, 27, 36], zero-shot learning [21,30,33], self-supervised learning [31], open-set learning [6, 22, 29] and advanced training strategies [17, 35] ”
- Towards Out-Of-Distribution Generalization: A Survey (2021, background): “Another related topic is open set classification, which aims to directly recognize unknown categories in test data [110, 79].”
- Federated learning for medical image analysis: A survey (2023, background): “, openset recognition (Geng et al., 2020; Qin et al., 2022)) will be a promising research direction.”
- A Model or 603 Exemplars: Towards Memory-Efficient Class-Incremental Learning (2022, background): “In the open world, training data is often collected in stream format with new classes appearing [16, 15, 63, 65].”
- Learning to Classify With Incremental New Class (2021, neutral): “As a result, images of these novel classes will be misclassified by the system, and classification performance will suffer obvious decay.”
- Learning Bounds for Open-Set Learning (2021, background): “This result reveals OSL problem can achieve an order of estimation error that is the same as CSL (Shalev-Shwartz & Ben-David, 2014).”
- Uncertainty-Aware Learning for Zero-Shot Semantic Segmentation (2020, background): “Zero-shot learning is a highly active research area in computer vision and machine learning [21, 56].”
- Domain Generalization: A Survey (2021, methodology): “…the model on the new dataset without training from scratch using all available datasets, ii) how to make sure the model does not over-fit the new dataset and forget the previously learned knowledge, and iii) will the new dataset be beneficial or detrimental to the DG performance on the target domain.”