[360Labs.ai]
{ 360Labs }Case Study
{ Hardware and IoT }· Live

CastVision

Automated quality inspection system for automotive manufacturing using X-ray imaging and deep learning to detect internal flaws in aluminum castings in real-time on the production line.

CastVision
// overview

X-ray image acquisition and preprocessing pipeline for aluminum casting inspection. Multi-model evaluation with synthetic data augmentation via 3D ellipsoidal simulation, eliminating manual annotation entirely.

// specs

Specifications

DetectionYOLOv3, YOLOv5, RetinaNet, EfficientDet
HardwareIndustrial X-ray tube (160kV-225kV), Jetson AGX Orin
Innovation3D ellipsoidal simulation for synthetic training defects
Inference18 FPS on-line GPU inference
// features

Features

  • 01Industrial X-ray tube with flat-panel digital detector for high-resolution imaging
  • 02Motorized rotary stage for multi-angle casting positioning and 3D volumetric capture
  • 03NVIDIA Jetson AGX Orin edge compute module for on-line GPU inference at 18 FPS
  • 04Conveyor-integrated proximity sensors and PLC triggers for automated image capture
  • 053D ellipsoidal simulation for synthetic training defect generation
// tech stack

Tech Stack

YOLOv3YOLOv5RetinaNetEfficientDetPyTorchX-Ray Imaging3D Simulation

Interested
in CastVision?

Let's discuss how this can work for you.

Get in Touch →
// end of case studyCastVision
Contact