01
The difficult problem
Internal casting defects required high-resolution capture and robust models that could run at line speed.
← All systemsVision & Edge AI
Industrial X-ray defect intelligence
CastScan
A real-time X-ray inspection system detecting internal flaws in aluminium castings on the production line.
VisionEdgeInfrastructureResearch
The brief
Built around the hard part.
02
The intelligence built
Multi-model defect detection combines X-ray preprocessing, synthetic training data and localized heatmaps.
System architecture
How the system works.
The model is one layer. The value comes from connecting inputs, intelligence and production action as one accountable system.
- 01
X-ray capture
Acquire and structure the operating signal.
- 02
3D defect simulation
Transform it through the model, rules and control layer.
- 03
Synthetic-trained inference
Deliver a decision, artifact or action into production.
In production
Where it runs.
Deployment
Industrial X-ray tube, flat-panel detector, rotary capture, conveyor trigger and Jetson AGX Orin.
What it changes
Detection and line rejection are connected in one edge inspection loop.
Production stack
- YOLO
- RetinaNet
- PyTorch
- TensorRT
- C++
- ROS