PBNA Utilities Asset Capture
Structured field evidence → Utilities Hub
Local OCR v0.6.4 • Parser Hotfix
ZERO PER-ASSET AI BILLING Multi-pass Tesseract OCR runs in this iPad/browser. Photos are not sent to OpenAI or another paid AI API by this app.
Starting local OCR… This build uses the same browser OCR family that already started successfully in v0.3, but adds multi-angle scans, spatial label/value matching, evidence scoring, and a real progress bar.
Starting… 0% • elapsed 0:00 • stage timeout 0:45
Field rule: the app must never make up a model, serial, year or rating. It reads the plate, keeps the text coordinates, matches labels to nearby values, and then runs a separate OEM date-recovery gate. Low-confidence fields are flagged for review.

1. Site and relationship

2. Capture field evidence

Nameplate photo: fill most of the frame with the plate, hold the iPad square to it, use enough light, and avoid direct glare. v0.6 automatically checks orientation, crops the detected text region, and runs a higher-detail contrast pass.

Overall asset photo Whole unit and surrounding context.
Nameplate / data plate Primary evidence for automatic extraction.
Wait for the green Local OCR ready banner.
Asset tag / QR / barcode Optional. The browser will scan supported codes locally.
Associated equipment Motor, separator, VFD, pump, controls, etc.

4. Extracted asset data

Green = strong read. Yellow = review carefully. Red = weak/uncertain. The visible OCR evidence is shown below so the auditor can verify each critical value before saving.

Electrical / mechanical

Process / capacity

Descriptive / other plate data

Preserved recognized nameplate data
Anything OCR reads but cannot safely map stays here. Low-confidence fragments remain in the raw OCR transcript; meaningful unmapped lines are surfaced here.
0 unmapped lines
Run a scan to populate preserved evidence.
Run a scan to evaluate completeness.
OCR evidence and field matching
Use this only if the auto-orientation picks the wrong direction.