SURFACE ANOMALY DETECTOR
The multi-camera solution scans up to 120 parts per minute and helps detect 90% of waste from defective parts and is able to classify parts by model and reject unknown parts. The single camera solution allows for faster inline inspection speeds (up to 800 pieces per minute) or the compactness to be mounted on a robotic arm for scanning larger surfaces.
Project: ZDZW
Publisher: UPV, VIDEO SYSTEMS, ITI
Technology: A camera-based inspection and AI algorithms
Industry: Manufacturers that need to ensure quality on their whole production
INTRODUCTION
The Surface Anomaly Detector Solution aims to use a camera-based inspection to analyze the surface of manufactured goods and detect visual defects like contamination, scratches, and roughness. By highlighting areas with defects, it prevents the inclusion of faulty pieces in the final product, reducing waste and energy consumption. AI will be utilized to train algorithms that can identify abnormal zones in the object by analyzing images.
FEATURES/BENEFITS
- Features:
- Inline inspection
- Captured object alignment
- Surface texture learning
- Surface anomaly detection
- Benefits
- Flexibility: Part classification, avoiding product specific solutions (part positioning and fixture exchange)
- Live monitoring of all production parts
- Cost reduction: No down-time (just loading calibration line) and no maintenance
- Millimetrical surface defects that human eye cannot detect
TECHNICAL INFORMATION
RESOURCES
SUCCESS STORIES:
- Bottle stoppers manufacturer with the next problems:
- Hundreds of references checked with human visual inspection
- Millimetrical surface defects not detected and sent to the clients
- With the Surface Anomaly Detector solution we solve the previous problem achieving these objectives:
- Reference classification and algorithms trained for each stopper/product
- High precision surface defects detection avoiding customer product return
Single camera and programmable multicolor illuminator 2D solution.Aero space mechanical parts inspection: https://youtu.be/sZMAubomiJE