Diffusion-based Image Generation for In-distribution Data Augmentation in Surface Defect Detection

TL;DR

This paper demonstrates that diffusion models can generate realistic in-distribution defect data for surface defect detection, improving data augmentation and classification performance in industrial scenarios.

Contribution

The study introduces a novel diffusion-based data augmentation method called In&Out, effective in zero-shot and few-shot defect scenarios, setting new state-of-the-art results.

Findings

Achieved a classification AP score of .782 on Kolektor Surface-Defect Dataset 2.

Diffusion models produce more realistic in-distribution defects than traditional superimposition methods.

The In&Out approach improves defect detection under weak supervision.

Abstract

In this study, we show that diffusion models can be used in industrial scenarios to improve the data augmentation procedure in the context of surface defect detection. In general, defect detection classifiers are trained on ground-truth data formed by normal samples (negative data) and samples with defects (positive data), where the latter are consistently fewer than normal samples. For these reasons, state-of-the-art data augmentation procedures add synthetic defect data by superimposing artifacts to normal samples. This leads to out-of-distribution augmented data so that the classification system learns what is not a normal sample but does not know what a defect really is. We show that diffusion models overcome this situation, providing more realistic in-distribution defects so that the model can learn the defect's genuine appearance. We propose a novel approach for data augmentation that mixes out-of-distribution with in-distribution samples, which we call In&Out. The approach can deal with two data augmentation setups: i) when no defects are available (zero-shot data augmentation) and ii) when defects are available, which can be in a small number (few-shot) or a large one (full-shot). We focus the experimental part on the most challenging benchmark in the state-of-the-art, i.e., the Kolektor Surface-Defect Dataset 2, defining the new state-of-the-art classification AP score under weak supervision of .782. The code is available at https://github.com/intelligolabs/in_and_out.