Contrastive Knowledge Distillation for Anomaly Detection in Multi-Illumination/Focus Display Images
Lists
Only you see this
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).
BibTeX
@inproceedings{lee2023contrastive,
author = {Jihyun Lee and Hangil Park and Yongmin Seo and Taewon Min and Joodong Yun and Jaewon Kim and Tae-Kyun Kim},
title = {{Contrastive Knowledge Distillation for Anomaly Detection in Multi-Illumination/Focus Display Images}},
booktitle = {MVA},
year = {2023},
url = {https://openreview.net/forum?id=3PI8WSiZ8Yv},
}Loading related content...