Manufacturing | Quality Assurance
Defect Detection to Enhance Quality Assurance
Objective
Manufacturing companies have immense Quality Assurance requirements and KPIs to detect faults/defects in their products and production lines. They also want to automate the entire QA process and reduce manual intervention.
Challenges
- Manual inspection is time-consuming and prone to human error.
- Limited sample size for defect detection.
- Need for real-time defect detection during production.
Solution Proposed
- Leveraged Deep Learning-based Convolutional Neural Networks (CNN) to classify and localize defects in fabric images.
- Implemented image augmentation techniques to boost the number of training images.
- Automated the QA process to reduce manual spot checks and improve accuracy.
Outcome
- Identified defect types and categories of target defects with 90% automation.
- Localized defects with utmost accuracy.
- Brought down manual QC time from days to real-time.
- Enhanced productivity and reduced operational costs.
Design & Thinking Wins
- Implemented the solution using an online tool through REST API.
- Achieved 80% accuracy in determining the instance and degree of corrosion.
- Reduced manual QC time significantly.
Disclaimer: The outline showcases the typical challenges, solutions, designs, and outcomes for industries and functions, in general, based on Affine’s prowess in the Industry. The outcomes would be much higher for specific clients as they would be based on their data and specific problems to be solved.
