Success case
Surface Crack Detection and Localization with Seeed’s Raspberry Pi-powered reTerminal
Concrete structures typically show signs of deterioration after 40 to 50 years, emphasizing the
- Industry
- Construction & Infrastructure
- Application
- Concrete surface cracks
Project background and needs
Japan-based engineer Naveen Kumar deployed Edge Impulse and a crack-detection model on reTerminal, using the portable device to support concrete structure inspection, crack localization, and data transfer.
Missed detections and consistency
- Manual visual inspection of large structures tends to miss small cracks.
- Cracks can be harder to observe when temperature changes cause material contraction.
Needs a crack-detection method that supports ongoing analysis.
On-site inspection workflow
- Traditional inspection relies on specialists observing and manually recording findings.
- This process is costly and time-consuming, with limited precision and consistency.
Needs a portable device that displays analysis results and supports copying and transferring data.
Solution and architecture
Model development
The developer selected a pre-trained image classification model in Edge Impulse Studio and fine-tuned it with transfer learning.
On-device application
The tuned model runs on reTerminal's Raspberry Pi CM4, detecting and localizing concrete surface cracks in real time.
Solution components
See the solution in action
Discuss your application with our team
Tell us about your requirements and deployment environment.