Case study
Computer Vision
Defect Detection
The Challenge
Heating, rolling, drying cutting – producing flat sheet steel is a delicate process and can lead to surface defects.
Detecting these – often very small – defects by visual inspection is time consuming and error prone.
The Solution
AUTOMATED DEFECT DETECTION WITH CONVOLUTIONAL NEURAL NETWORKS
1. Existing data from cameras in the production line is annotated
2. Data is preprocessed and used to train a Convolutional Neural Network
3. The algorithm is integrated into the production system
Benefits
- Higher detection rate compared to manual inspection
- Automated inspection speeds up the production process
- Extracted information, e.g. position of the defect, can be used in subsequent processing steps
Further Use Cases
- Package inspection: Detect missing elements like bar-codes or damages
- Healthcare: Analyse medical scans and detect diseases
- Agriculture: Detect pests and plant diseases
- Traffic: Count and classify road users
... and many more
Other Case Studies
Contact us