DC Field | Value | Language |
---|---|---|
dc.contributor.author | Park, Hangil | ko |
dc.contributor.author | Lee, Jihyun | ko |
dc.contributor.author | Seo, Yongmin | ko |
dc.contributor.author | Min, Taewon | ko |
dc.contributor.author | Yun, Judong | ko |
dc.contributor.author | Kim, Jaehwan | ko |
dc.contributor.author | Kim, Tae-Kyun | ko |
dc.date.accessioned | 2023-11-28T08:04:27Z | - |
dc.date.available | 2023-11-28T08:04:27Z | - |
dc.date.created | 2023-11-27 | - |
dc.date.issued | 2023-07-24 | - |
dc.identifier.citation | 18th International Conference on Machine Vision and Applications, MVA 2023 | - |
dc.identifier.uri | http://hdl.handle.net/10203/315354 | - |
dc.description.abstract | In this paper, we tackle automatic anomaly detection in multi-illumination and multi-focus display images. The minute defects on the display surface are hard to spot out in RGB images and by a model trained with only normal data. To address this, we propose a novel contrastive learning scheme for knowledge distillation-based anomaly detection. In our framework, Multiresolution Knowledge Distillation (MKD) is adopted as a baseline, which operates by measuring feature similarities between the teacher and student networks. Based on MKD, we propose a novel contrastive learning method, namely Multiresolution Contrastive Distillation (MCD), which does not require positive/negative pairs with an anchor but operates by pulling/pushing the distance between the teacher and student features. Furthermore, we propose the blending module that transforms and aggregate multi-channel information to the three-channel input layer of MCD. Our proposed method significantly outperforms competitive state-of-the-art methods in both AUROC and accuracy metrics on the collected Multi-illumination and Multi-focus display image dataset for Anomaly Detection (MMdAD). | - |
dc.language | English | - |
dc.publisher | Institute of Electrical and Electronics Engineers Inc. | - |
dc.title | Contrastive Knowledge Distillation for Anomaly Detection in Multi-Illumination/Focus Display Images | - |
dc.type | Conference | - |
dc.identifier.wosid | 001057888900029 | - |
dc.identifier.scopusid | 2-s2.0-85170521491 | - |
dc.type.rims | CONF | - |
dc.citation.publicationname | 18th International Conference on Machine Vision and Applications, MVA 2023 | - |
dc.identifier.conferencecountry | JA | - |
dc.identifier.conferencelocation | Hamamatsu | - |
dc.identifier.doi | 10.23919/MVA57639.2023.10215808 | - |
dc.contributor.localauthor | Kim, Tae-Kyun | - |
dc.contributor.nonIdAuthor | Seo, Yongmin | - |
dc.contributor.nonIdAuthor | Min, Taewon | - |
dc.contributor.nonIdAuthor | Yun, Judong | - |
dc.contributor.nonIdAuthor | Kim, Jaehwan | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.