Defect Spectrum: A Granular Look of Large-Scale Defect Datasets with Rich Semantics

TL;DR

The paper introduces the Defect Spectrum dataset with rich semantic annotations for industrial defects and a diffusion-based generator, Defect-Gen, to produce diverse synthetic defective images, advancing defect inspection research.

Contribution

It provides a large-scale, semantically detailed defect dataset and a novel generative model to improve defect inspection methods.

Findings

Enhanced defect detection accuracy with the dataset.

Synthetic images improve model robustness.

Rich semantic annotations enable finer defect differentiation.

Abstract

Defect inspection is paramount within the closed-loop manufacturing system. However, existing datasets for defect inspection often lack precision and semantic granularity required for practical applications. In this paper, we introduce the Defect Spectrum, a comprehensive benchmark that offers precise, semantic-abundant, and large-scale annotations for a wide range of industrial defects. Building on four key industrial benchmarks, our dataset refines existing annotations and introduces rich semantic details, distinguishing multiple defect types within a single image. Furthermore, we introduce Defect-Gen, a two-stage diffusion-based generator designed to create high-quality and diverse defective images, even when working with limited datasets. The synthetic images generated by Defect-Gen significantly enhance the efficacy of defect inspection models. Overall, The Defect Spectrum dataset demonstrates its potential in defect inspection research, offering a solid platform for testing and refining advanced models.