KAORI MODEL SUITE V2.5
Kaori Visual Intelligence
Zero-shot document spatial understanding, OCR, layout extraction, and Indian manuscript paper texture parsing.
05 — KAORI VISUAL INTELLIGENCE
KAORI
Visual spatial understanding & document OCR.
Kaori sees documents as organic visual surfaces. Inspired by Indian botanical textures and manuscript geometry, Kaori parses dense multi-column research, forms, and historical documents into clean structured schemas.
DOCUMENT_OCR_STREAM // SCAN #8941
CONFIDENCE: 99.82%TITLE_NODE [H1]
“A Spatial Study on Indian Botanical Manuscripts & Vector Layout Alignment”
PARAGRAPH_LEFT
MANUSCRIPT_GRID_SCHEMA
{ "layout": "grid_2col", "entities": 14, "script": "latin_sanskrit" }JSON EXPAND →
BOTANICAL & FLORAL SURFACE SENSING
Zero-shot table & form structure extraction
Historical manuscript & multilingual Sanskrit/Hindi OCR support
Spatial coordinate layout synthesis for PDF pipelines
Layout Synthesis Engine
Maps multi-column research papers, tables, formulas, and historical Indian manuscripts into clean JSON schemas.
Sub-15ms Latency
Optimized for high-throughput enterprise document pipelines with FP8 quantized model execution.
Multilingual OCR
Supports Devanagari, Sanskrit, Latin, Arabic, and 24 global languages out of the box.