Foil Alignment Vision System (FAVS)scalable inspection platform
The Challenge#
- Packaging defects are among the top causes of product rejection and recalls
- Scrap rates in packaging processes can range from 2% to 8%, depending on process stability
- Manual inspection introduces inconsistency and limited traceability
- Many defects originate upstream but are only detected late in the process
They need real-time, inline inspection that not only detects defects—but helps prevent them.
- Inconsistent alignment of sealing material across cavities
- No reliable way to quantify alignment deviations in real time
- Dependence on manual or downstream inspection
- Increasing scrap and rework costs
- Requirement to maintain strict compliance and traceability
- Inspect at full production speed
- Provide measurable, actionable data
- Integrate seamlessly into their existing line
- Enable immediate process correction
The Solution: FAVS (Foil Alignment Vision System)#
custom inline vision systemApplication Scope
- Inspection of multi-unit arrays (15 units per cycle)
- Continuous inline operation without slowing production
- Real-time pass/fail decisioning and rejection
Core Capabilities
- Detection of misalignment in both horizontal and vertical directions
- Evaluation at unit-level and array-level
- Automatic rejection if any unit fails
- Continuous data logging and statistical tracking
- Feedback loop to support upstream process adjustments
reactive inspection → proactive control
System Architecture #
high-performance, modular vision platformHardware Components
- Lenovo Industrial PC
- Equipped with NVIDIA GPU
- StarTech GigE Ethernet card for high-speed camera connectivity
- Machine Vision Module (MVM):
- 3 × FLIR Blackfly cameras
- Each camera captures a 5-unit array simultaneously
- Advanced Illumination System:
- Brightfield lighting for structural visibility
- Darkfield lighting for edge and defect contrast
- Dual light controllers for precision tuning
Software & Platform: EZ Eye FrameworkEZ Eye Framework
- Synchronized multi-camera image acquisition
- Real-time inspection processing
- Pass/fail decisioning logic
- Data logging for traceability
- Statistical analysis for trend monitoring
multiple inspection methodologiesInspection Methodology & Algorithm Designdeterministic computer vision approachCognex VisionProWhy This Approach Works
- High-speed performance suitable for inline inspection
- Fully explainable and traceable results
- No dependency on large training datasets
- Ideal for regulated manufacturing environments
Inspection Workflow
- Package Localization
- Each camera has a unique view of the production line
- Custom detection logic identifies package positions accurately
- Alignment Measurement
- Packages act as reference geometry
- System calculates:
- Vertical displacement (up/down)
- Horizontal displacement (left/right)
- A midpoint reference is used to determine alignment deviation
- Decision Logic
- Each unit is evaluated against tolerance thresholds
- Any failure triggers rejection of the full array
Operator Experiencefaster root cause identification by operators
- Real-time defect overlays highlighting failure locations
- Toggle between clean and diagnostic views
Production Results & Impact#
Quality Improvements
- Early detection of defects before downstream processing
- Improved packaging consistency and reliability
- Strengthened regulatory compliance
Operational Gains
- Significant reduction in scrap through immediate feedback
- Real-time adjustments to upstream processes
- Elimination of manual inspection variability
Data-Driven Insights
- Continuous monitoring of alignment trends
- Identification of process drift
- Support for continuous improvement initiatives
From Customer Solution to Scalable Platform#
repeatable and adaptable systemApplicable Use Cases
- Blister pack alignment verification
- Seal placement inspection
- Film or lid positioning
- Multi-cavity packaging inspection
- Any application requiring relative position measurement between components
Platform Advantages
- Hardware-agnostic design
- Modular architecture via EZ Eye Framework
- Supports:
- Rules-based vision
- Machine learning models
- Hybrid inspection strategies
- Improve quality
- Reduce waste
- Gain deeper visibility into their manufacturing process
flexible platform