Project Overview
QualityAI, a electronics manufacturer, wanted to automate visual inspection of circuit boards. Intelirate deployed a YOLOv8‑based computer vision system.
Services Provided
- Custom dataset annotation (20k images)
- Object detection for 15 defect types (solder, scratches, missing components)
- Real‑time inference on edge devices (NVIDIA Jetson)
- Reject mechanism integration
- Dashboard with defect heatmaps
Results Achieved
- 98% defect detection accuracy (vs 85% human)
- Inspection speed 200 units/minute (5x faster)
- False positives reduced by 70%
- Saved $1.2M/year in rework and returns
Client Feedback
“Our quality control is now faster and more accurate than any human team. Intelirate’s vision system paid for itself in 4 months.”
Tech Stack
- Python, YOLOv8, OpenCV
- NVIDIA Jetson, TensorRT
- LabelImg, Roboflow