Visual Quality Inspection System
Impact99.2% defect detection
Detection Rate99.2%
Speed60 FPS
False Positives<1%
Overview
Developed an automated visual inspection system for a manufacturing company to detect product defects in real-time on the production line. The system uses deep learning models optimized for edge deployment.
The Challenge
Manual quality inspection was slow and inconsistent, with defects often slipping through, leading to customer complaints and recalls.
The Solution
Trained YOLO-based defect detection models on labeled defect images, optimized for NVIDIA Jetson edge deployment. Built integration with PLC systems for automatic rejection of defective items.
Key Results
99.2% defect detection rate
Processing at 60 FPS
<1% false positive rate
Tech Stack
PythonPyTorchYOLOv8NVIDIA JetsonTensorRTOpenCVMQTT