Beyond Barcodes:
Custom AI Vision
Tailored to Your Business
Case1Read test tube barcodes while checking cap presence.
Case2Decode barcodes while recognizing printed text.
What once required two devices now fits in one.
MDI-4700 / MDI-5350 / NLV-5201S/W (Coming Soon) ／ Edge AI Barcode Reader
NLV-5201S/W (Coming Soon)
Processor Built-in
Edge Inference
640 × 480
1280 × 800
Close-range
AI OCR
No PC needed
Main Visual Image (1180 × 300 px)
Replace with src="/dcms_media/image/lp-images/fv_main.jpg?02"
Are These Challenges Holding You Back?
In manufacturing, logistics, and medical environments, barcode reading alone can no longer meet growing inspection demands.
Barcode reading and visual inspection
handled by separate systems
You need to read test tube barcodes while also verifying cap presence and type. But today you run a separate vision camera alongside the barcode reader — adding cost, space, and wiring complexity.
A missed barcode scan means
manual visual re-check
Peeling labels or poor print quality stop the line and require manual verification. Items without barcodes also cause sorting errors that are costly to catch and correct.
Label and print verification
still relies on human eyes
Expiry dates, lot numbers, and label placement are checked manually. This creates ongoing labor costs and quality inconsistency, making it difficult to consolidate inspection equipment.
MDI-4700 / MDI-5350 / NLV-5201S/W (Coming Soon) solves all of these challenges in one device
4 AI Solutions
MDI-4700 / MDI-5350 / NLV-5201S/W (Coming Soon) runs AI processing simultaneously with barcode reading. Four solutions tailored to your application.
Object Detection
Barcode + AI Detection
While decoding barcodes, the AI simultaneously inspects the object's appearance. Cap presence, label attachment, mark detection — checks that once required a separate camera are now done by one device.
An optimized, lightweight deep learning engine runs on-device using the built-in AI inference processor — no external host required.
AI OCR
Text Recognition & Barcode Fallback
Linked to barcode scan results, the AI recognizes printed text on labels. When a barcode scan fails, text recognition acts as an automatic fallback, keeping the line running.
Expiry dates, lot numbers, product names — custom tuning adapts to each customer's label format.
AI Count
Quantity Verification & NG Detection
The AI automatically detects and counts objects in the camera's field of view. "How many good parts?" "How many defects?" Real-time counting happens simultaneously with scanning.
Automates manual counting tasks: production line quantity checks, pre-shipment verification, and inventory stock counts.
Custom Tuning
Optimized for Your Environment
For cases where the standard model falls short, we retrain the AI using your actual production images and OK/NG samples — custom-tuned to your specific use case.
Optimized for your lighting conditions, read distance, and object size, enabling broad applicability across diverse deployment environments.
4 Key Strengths of MDI-4700 / MDI-5350 / NLV-5201S/W (Coming Soon)
The unique positioning of MDI-4700 / MDI-5350 / NLV-5201S/W (Coming Soon) compared to competing solutions.
Integrated Barcode Reading + AI Vision in One Device
Barcode decoding and AI image recognition run simultaneously in a single device. Where competing systems require separate products for barcode reading and visual inspection, MDI-4700 / MDI-5350 / NLV-5201S/W (Coming Soon) integrates both — dramatically reducing hardware cost, installation space, and wiring.
Compact Design Optimized for 50–500 mm Close Range
Designed to operate at a 50–500 mm working distance in a compact form factor. Easy to embed in test tube processing machines, conveyor lines, and collection bins. Enables space-saving installations that competing products cannot achieve.
Edge AI — No PC or Server Required
The MDI-4700 / MDI-5350 / NLV-5201S/W (Coming Soon) with its built-in NPU completes AI inference entirely on-device. No host PC or cloud server needed for operation, reducing deployment cost, IT management overhead, and communication failure risk. Existing NLV-5201 users can upgrade with a direct drop-in replacement.
Custom Tuning Tailored to Your Environment
AI models are tuned using your actual site images and OK/NG samples. Handles lighting conditions, object shapes, and read distances that standard models struggle with — delivering optimized performance for each customer's specific requirements.
Industry Use Cases
See the Before / After impact MDI-4700 / MDI-5350 / NLV-5201S/W (Coming Soon) delivers across industries.
Barcode Scanning + Cap Verification
Completed by One Device
- Separate barcode reader and vision camera installed
- Risk of uncapped test tubes being loaded incorrectly
- Large equipment footprint limits installation options
- Dual-system maintenance and management costs
- Barcode reading + cap detection handled by one device
- Incorrect loading detected automatically with alarm output
- Compact design integrates easily into existing equipment
- Consolidated system reduces management cost
- Reduced hardware cost by eliminating separate vision cameras
- Prevents damage to instruments and equipment
- Improves operator safety
- Compact design increases installation flexibility
Test Tube Barcode Reading
+ AI Cap Detection
(Product image / demo video coming soon)
AI OCR Keeps the Line Moving
Even When Barcode Scanning Fails
- Peeling labels or poor print quality causes line stops
- Items without barcodes require manual re-check
- PC-dependent systems drive up deployment costs
- Mis-sorting causes returns and customer complaints
- AI OCR automatically falls back to text recognition
- AI detects missing labels and alerts immediately
- Compact design retrofits easily onto conveyors
- Edge processing — no PC required
- Fewer line stops from unreadable barcodes
- Reduced labor cost for manual barcode re-checks
- Low-cost, PC-free equipment addition
Conveyor Line Installation
Barcode Reading + AI OCR Fallback
(Product image / demo video coming soon)
AI Automatically Counts Production Units & Defects
Eliminate Miscounts Forever
- Production and shipment quantities counted manually
- Tracking defect counts is slow and error-prone
- Quantity discrepancies discovered after packaging cause rework
- Inventory counts require significant manual labor
- AI counts good parts and defects in real time
- Alarm triggered immediately if count differs from target
- Count results automatically linked to barcode data and logged
- Inventory counts automated to reduce labor hours
- Fewer reworks and customer complaints from quantity errors
- Prevents missed defects and automates quality records
- Reduced labor hours for inventory and quantity checks
Production Line & Inventory
AI Count + Barcode Reading
(Product image / demo video coming soon)
Label Verification, Print Inspection & Barcode Reading
Completed in a Single Pass
- Expiry dates and lot numbers verified by manual visual inspection
- Risk of missing labels going undetected
- Barcode scanning and print inspection as separate processes
- Multiple devices mean high maintenance costs
- AI OCR automatically reads expiry dates and lot numbers
- Object detection auto-checks label presence
- Completed simultaneously with barcode scan (single-pass inspection)
- Multiple inspections consolidated into one device
- Fewer manual inspection staff needed; stable quality
- Reduced risk of label defects reaching customers
- Lower equipment costs through inspection consolidation
Food Label Inspection Line
Barcode + OCR + Label Presence Check
(Product image / demo video coming soon)
Automate Deposit Refunds
with Recycling Mark Recognition
- Material types (aluminum, steel, etc.) identified manually
- High-cost AI cameras impractical to deploy in quantity
- PC-based system maintenance and management costs
- Human sorting errors cause contamination risk
- Recycling mark recognition AI automates deposit refund processing
- Compact and affordable enough for multi-unit deployment
- Edge AI processing — standalone, no PC needed
- Multi-angle recognition improves detection accuracy
- Reduced labor cost for sorting operations
- Improved recognition coverage with multi-unit setup
- Lower maintenance cost with PC-free operation
Collection Bin — Multi-Unit Setup
Recycling Mark AI Recognition
(Product image / demo video coming soon)
Product Specifications
Hardware and AI performance specifications for MDI-4700 / MDI-5350 / NLV-5201S/W (Coming Soon).
| Processor | Embedded Processor + Built-in AI Inference Processor |
|---|---|
| CMOS Sensor | Industrial Global Shutter Monochrome Sensor 640 × 480 1280 × 800 |
| Read Range | 50 – 500 mm (close-range optimized) |
| Illumination | Designed for close-range operation (recommended lighting settings available) |
| OS | Real-Time OS |
| AI Processing | Edge Inference Engine (fully on-device) |
| PC / Server | Not required for inference (edge-complete) PC/GPU environment needed for training only |
| Object Detection Accuracy | To be discussed (varies by environment and conditions) |
|---|---|
| AI OCR Accuracy | To be discussed (varies by environment and conditions) |
| Inference Time | To be discussed (varies by subject and conditions) |
| False Positive Rate | To be discussed (varies by environment and conditions) |
| AI Engine | Deep Learning Engine (object detection) |
| AI OCR Engine | Proprietary AI-OCR Engine |
| Custom Support | Additional training and tuning with customer images available |
| Supported Characters | Numerals, Latin alphabet (Japanese: phased rollout planned) |
* Custom tuning can optimize accuracy for your specific environment.
* Next-generation model (enhanced AI inference processor + 1M CMOS sensor) is under development.
Contact Us / Request Information
We'd love to hear from you. Reach out anytime.
Demo units and on-site consultations are available.