Search:
Match:
1 results

Analysis

This paper addresses the challenge of anomaly detection in industrial manufacturing, where real defect images are scarce. It proposes a novel framework to generate high-quality synthetic defect images by combining a text-guided image-to-image translation model and an image retrieval model. The two-stage training strategy further enhances performance by leveraging both rule-based and generative model-based synthesis. This approach offers a cost-effective solution to improve anomaly detection accuracy.
Reference

The paper introduces a novel framework that leverages a pre-trained text-guided image-to-image translation model and image retrieval model to efficiently generate synthetic defect images.