UC III · Camera AI
Camera AI
Document OCR (Form-13 / EIR / PIN), gate object detection (person · car · truck · plate · trailer · container), and PDF report parser (berthing / daily status / TEUs — detects Excel vs Jasper origin).
Preview · crop before OCR
Drag the box to select the document area. OCR uses only the cropped region.
No image selected
Result summary
Raw OCR text (PaddleOCR)
(none)
Extracted fields (LLM rearrange)
{}
Layout pairs (box-based)
{}
Full JSON response
{}
Vision OCR via Ollama qwen2.5vl:7b — image in, raw text out. Current Document OCR (PaddleOCR + qwen3) is unchanged.
Preview
No image selected
Summary
Classification
{}
Raw OCR text (qwen2.5vl)
(none)
Full JSON response
{}
Annotated result
No detection yet
Summary
Vehicles
[]
Container OCR (aggregated)
{}
Result JSON
{}
InsightFace buffalo_l — match against
face_data/embeddings. Unknown faces auto-register when enabled.
Use File upload or live Camera.
Live camera needs HTTPS or localhost. On HTTP LAN IP, use Take photo instead.
Preview
No file selected
Summary
Gallery / people
[]
JSON response
{}
Detection (how was this PDF made?)
Parsed data (preview)
{}
Full JSON response
{}