Related Experiment Video
Updated: Sep 14, 2025
07:46
Data Acquisition Protocol for Determining Embedded Sensitivity Functions
Published on: April 20, 2016
6.2K
BladeSynth: A High-Quality Rendering-Based Synthetic Dataset for Aero Engine Blade Defect Inspection
M A Mohammed Eltoum1, Ehtesham Iqbal1, Yahya Zweiri1,2
1Advanced Research and Innovation Center (ARIC), Khalifa University of Science and Technology, Abu Dhabi, United Arab Emirates.
Scientific Data
|July 19, 2025
Summary
Generating synthetic aeroengine blade data using physics-based rendering addresses industrial data scarcity. This approach enhances defect detection accuracy for Industry 4.0 applications.
Area of Science:
- Engineering
- Computer Science
Background:
- Industry 4.0 integration relies on artificial intelligence (AI), but industrial datasets are scarce.
- Existing generative AI methods for synthetic data are often inefficient and data-hungry.
Purpose of the Study:
- To develop an efficient method for generating synthetic aeroengine blade datasets.
- To address data scarcity challenges in industrial AI applications.
- To improve defect detection accuracy using synthetic data.
Main Methods:
- Utilized a physics-based rendering procedure for synthetic dataset generation.
- Prepared Computer-Aided Design (CAD) models and material textures.
- Constructed realistic inspection scenes with domain-randomized parameters (camera, lighting, background).
Main Results:
- Generated a synthetic dataset of aeroengine blades.
- Trained a defect inspection model using the synthetic dataset.
- Demonstrated effectiveness in both supervised and unsupervised defect detection tasks.
- Validated sim-to-real transferability for real-world defect classification.
Conclusions:
- Physics-based rendering is an effective method for generating industrial synthetic data.
- Synthetic data significantly enhances defect detection accuracy.
- Models trained on synthetic data exhibit strong performance on real-world industrial inspection tasks.
