Defect Detection

Overview

According to the American Society of Quality (ASQ), defects cost manufacturers nearly 20% of overall sales revenue. However, this revenue loss is an opportunity for toolmakers; in both manufacturing and field inspection scenarios, AI-based computer vision applications can help catch defects much faster and more effectively than traditional inspection methods, allowing companies to increase yield, deliver products with consistent quality, and reduce false positives. However, training such perception AI models requires collecting images of specific defects, which is often difficult and expensive to do in a production environment.

The Omniverse Platform can help overcome the data challenge by generating synthetic data to bootstrap the AI model training process using Omniverse Replicator, which is an extensible foundation application in Omniverse.

This guide will point you to resources that will allow you to build your own tools to easily create synthetic data for your use case.

Download and Install the Extension

Download and the Defects Generation Extension from GitHub

Download the Content

Search “defect detection” in Omniverse Exchange to find the content used in the scratch detection sample.

Workflow Steps

• Import Scene Data
• Build a Scene
• Generate Synthetic Data
• Model Training

Note

Before you begin, install Omniverse Code version 2022.3.1 or higher.