Calibrated Out-of-Distribution Detection with a Generic Representation

被引:2
作者
Vojir, Tomas [1 ]
Sochman, Jan [1 ]
Aljundi, Rahaf [2 ]
Matas, Jiri [1 ]
机构
[1] Czech Tech Univ, CMP Visual Recognit Grp, FEE, Prague, Czech Republic
[2] Toyota Motor Europe, Toyota, Belgium
来源
2023 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION WORKSHOPS, ICCVW | 2023年
关键词
D O I
10.1109/ICCVW60793.2023.00485
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Out-of-distribution detection is a common issue in deploying vision models in practice and solving it is an essential building block in safety critical applications. Most of the existing OOD detection solutions focus on improving the OOD robustness of a classification model trained exclusively on in-distribution (ID) data. In this work, we take a different approach and propose to leverage generic pre-trained representation. We propose a novel OOD method, called GROOD, that formulates the OOD detection as a Neyman-Pearson task with well calibrated scores and which achieves excellent performance, predicated by the use of a good generic representation. Only a trivial training process is required for adapting GROOD to a particular problem. The method is simple, general, efficient, calibrated and with only a few hyper-parameters. The method achieves state-of-the-art performance on a number of OOD benchmarks, reaching near perfect performance on several of them. The source code is available at https://github.com/vojirt/GROOD.
引用
收藏
页码:4509 / 4518
页数:10
相关论文
共 52 条
[1]  
[Anonymous], 2022, PR MACH LEARN RES
[2]   The MVTec Anomaly Detection Dataset: A Comprehensive Real-World Dataset for Unsupervised Anomaly Detection [J].
Bergmann, Paul ;
Batzner, Kilian ;
Fauser, Michael ;
Sattlegger, David ;
Steger, Carsten .
INTERNATIONAL JOURNAL OF COMPUTER VISION, 2021, 129 (04) :1038-1059
[3]  
Bitterwolf Julian, 2023, ICLR 2023 WORKSH PIT
[4]  
Bitterwolf Julian, 2022, BREAKING OUT DISTRIB
[5]  
Bogdoll Daniel, 2022, IEEE C COMP VIS PATT
[6]  
Chan Robin, 2021, SegmentMeIfYouCan: A Benchmark for Anomaly Segmentation
[7]   On the Effect of Heat Treatments on the Adhesion, Finishing and Decay Resistance of Japanese cedar (Cryptomeria japonica D. Don) and Formosa acacia (Acacia confuse Merr.(Leguminosae)) [J].
Chang, Chia-Wei ;
Kuo, Wei-Ling ;
Lu, Kun-Tsung .
FORESTS, 2019, 10 (07)
[8]   Ridgelet moment invariants for robust pattern recognition [J].
Chen, Guang Yi ;
Li, Changjun .
PATTERN ANALYSIS AND APPLICATIONS, 2021, 24 (03) :1367-1376
[9]   Describing Textures in the Wild [J].
Cimpoi, Mircea ;
Maji, Subhransu ;
Kokkinos, Iasonas ;
Mohamed, Sammy ;
Vedaldi, Andrea .
2014 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2014, :3606-3613
[10]  
Dhamija Akshay Raj, 2018, ADV NEURAL INFORM PR