PROJECT TOPICComputer Vision & Embedded AI
Optical Character Recognition Picker
A parcel handling concept exploring OCR neural pipelines capable of parsing printed addresses, serial numbers, and postal codes directly from parcel surfaces to trigger routing diversions.
Text-Based Parcel Logistics Automation
- ✓Applies deep text detection (EAST/DBNet) combined with text recognition models.
- ✓Parses multi-oriented text on rumpled, taped, or non-planar parcel surfaces.
- ✓Triggers high-speed pneumatic divert flippers within 200 milliseconds of text capture.
The Engineering Challenge
Problem / Objective
Handwritten or non-standard parcel shipping labels cannot be read by conventional barcode scanners, requiring costly manual sorting lanes.
Conceptual Signal Flow
System Concept
Surface Lighting → High-Res Camera → Text Detection & Recognition Engine → Diverter Conveyor.
Architecture Modules
1Overhead Ring Illuminator & Camera
2Edge Tensor Text Extraction Unit
3Fuzzy Text Matching & Postal Parser
4Diverter Conveyor Interface
Hardware Categories
High-Res Global Shutter CameraLED Ring IlluminatorEdge AI GPU BoardPneumatic Diverter
Technologies & Software
OCRDeep LearningParcel LogisticsMachine VisionTesseract / PaddleOCR PipelinePostal Routing Database Matcher
Engineering Considerations & Edge Cases
- Text distortion on curved envelopes or reflective clear tape
- Latency required for items moving at 2 meters per second
Applications
Postal sorting facilities
E-commerce return processing
Pharmaceutical label verification
Engineering Inquiry & Collaboration
Interested in Developing this Architecture?
Connect directly with Tamizh Tech engineers in Coimbatore to discuss mechanical fabrication, sensor selection, firmware implementation, or turnkey system commissioning.