iTromsø · IBM, VC2
10 minutes per day — time spent reviewing potential stories
Independently reportedMedia & entertainment · Document processing · NLP · In production
Document processing
Fleetoptics · Jamil Global
Vendor-reportedLogistics & transportIn productionComputer vision
Vendor-reported. The customer is named and the numbers are quoted from the source page, but the account comes from the vendor. No independent confirmation.
Reported by jamilglobal.com (vendor self-report). Checked against the source page on 2026-09-06. 2 of 2 figures below appear on that page word for word.
The problem
Fleetoptics needed a reliable, automated way to verify delivery addresses from phone photos taken in unpredictable field conditions to prevent mis-deliveries at scale. Manual checks were slow and inconsistent.
What was deployed
- Vendor
- Jamil Global
- Models
- YOLO, PaddleOCR, Google OCR, Tesseract
- Technique
- Computer vision
- Build or buy
- Built in-house or to order
- Deployment
- API
- Data used
- parcel photos directly from delivery sites, real-world delivery imagery, phone photos taken in unpredictable field conditions
- Scale
- 15,000+ images validated
- Timeline
- 2 months
What changed
Each row is quoted from the source. Figures we could not find on the page in those words are marked — they are kept, not deleted, so you can judge them.
77.4Address verification accuracy
“77.4% — Address verification accuracy”
Quoted word for word from the source
15000+Images validated
“15,000+ — Images validated in production runs”
Quoted word for word from the source
Dependence on manual Virtual Assistant review
“the flow reduced dependence on manual Virtual Assistant review”
Quoted word for word from the source
Address verification speed
“Address verification became much faster”
Quoted word for word from the source
Address verification dependability
“far more dependable”
Quoted word for word from the source
Portion of delivery address checks automated
“The platform automated a large portion of delivery address checks that were previously manual and delayed final delivery closure.”
Quoted word for word from the source
Driver workflow speed
“Drivers moved through delivery workflows faster with immediate confidence checks available from AI validation”
Quoted word for word from the source
False misses and false matches
“The custom ensemble model architecture reduced false misses and false matches”
Quoted word for word from the source
“Address verification became much faster and far more dependable once the custom OCR ensemble was deployed across low-quality delivery photos.”
Difficulties and limits
- Drivers took parcel photos with Android and iPhone devices in poor lighting, extreme angles, and motion blur.
- Manual address checks were slow, inconsistent, and resource-heavy.
- Android/iPhone images were heavily compressed, under-processed, blurry, overexposed, and often clipped, causing severe OCR drift in a traditional pipeline.
- Varying labels, hand-written additions, font styles, and sticker placements required a defense-in-depth approach to extraction.
- Single OCR methods underperformed on inconsistent imagery, low-res captures, and mixed language/label styles from driver phone photos.
Sources
- jamilglobal.com — vendor self-report
Similar deployments
Amplitude · Wolfia
4X — throughput of security questionnaires
Vendor-reportedTechnology & software · Document processing · AI agents · Scaled
Adyen · Rossum
70 — automation rate
Vendor-reportedBanking & finance · Document processing · NLP · Scaled · 2023
Skaylink · Moterra
dropped by around 80% — NIS2 preparation time
Vendor-reportedTechnology & software · Document processing · LLM · Scaled
OM Pharma · DrugCard
3 minutes — review time per article
Vendor-reportedPharma & biotech · Document processing · NLP · In production
AMPREP (Amalgamations Repco) · Cimplify.AI
100% — invoice automation
Vendor-reportedManufacturing · Document processing · NLP · In production