Contextual candidate classification on a changing tobacco belt
Belt colour and conveyor cleats in a tobacco flow can resemble foreign-material candidates. Fixed blind regions can instead exclude real foreign objects passing through those locations.
Proposed output: candidate region, context class, visibility and review reason. Results require measurement by foreign-material type and size group; this scene does not show an accuracy percentage.
- 01 Presentation
- 02 Trigger and reference
- 03 Image / profile
- 04 Decision
- 05 Evidence and PLC
Scan the visible field for candidates
- Sample record
- —
- Context and visibility consistency
- —
- Acquisition progress
- —
Scenes are illustrative simulations; they are not customer references or performance guarantees.
Solution approach
Detect candidates across the visible field and classify each with its tobacco, belt and cleat context in a second stage. Track visibility and uncertainty rather than creating fixed blind regions. Revalidate with independent samples after belt changes; fully covered objects have no optical evidence.
Process steps
- 01
Presentation
Scan the visible field for candidates
- 02
Trigger and reference
Add surrounding context
- 03
Image / profile
Separate foreign-material and background classes
- 04
Decision
Retain uncertainty with visibility
- 05
Evidence and PLC
Record evidence by type and size
Related applications
Let us assess your application together
We test your sample in our laboratory and report on resolution, optics and lighting selection.