メインコンテンツへスキップ
📄 ドキュメント解析ファミリー

ドキュメント解析と レイアウトOCR

PDF、スキャン書類、請求書、契約書を構造化JSONやLLM RAGに最適なMarkdownへ高精度変換。

Document AI・レイアウトOCR

段組みや傾きのある複雑な文書をピクセル単位のバウンディングボックス付きで正確にテキスト認識。

  • Layout-aware text recognition with pixel-accurate bounding boxes
  • Multi-column, complex diagram, and rotated document parsing
  • Supports PDF, TIFF, PNG, JPEG with automatic skew correction
Page Speed< 350ms / page
Character Accuracy99.4% on printed text
POST /v1/document/extractcURL
curl -X POST https://api.brainiall.com/v1/document/extract \
  -H "Authorization: Bearer $BRAINIALL_KEY" \
  -F "file=@contract.pdf" \
  -F "features=layout,text,boxes"

# Response:
# {
#   "pages": [{
#     "page": 1,
#     "blocks": [{ "type": "header", "text": "Service Agreement", "box": [40, 50, 520, 90] }]
#   }]
# }

PDFからクリーンMarkdownへの変換

複雑なPDFや表組みをベクトル検索に適したセマンティックなGitHub Markdownに変換します。

  • Converts arbitrary PDFs into clean, semantic GitHub-flavored Markdown
  • Reconstructs nested tables, lists, headers, and code snippets faithfully
  • Optimized as the ingestion standard for LLM RAG pipelines
POST /v1/document/pdf-to-markdowncURL
curl -X POST https://api.brainiall.com/v1/document/pdf-to-markdown \
  -H "Authorization: Bearer $BRAINIALL_KEY" \
  -F "file=@annual_report.pdf" \
  -F "table_format=gfm"

# Returns pure semantic markdown ready for vector chunking:
# # Annual Financial Report 2026
# | Quarter | Revenue | EBITDA |
# |---|---|---|
# | Q1 | $14.2M | $3.8M |

構造化フォーム・表抽出

請求書や領収書からキー・バリューペアおよび表マトリックスを自動抽出。

  • Key-value pair extraction from unstructured receipts, bills, and forms
  • Reconstructs cell-by-cell table matrices with merged cell detection
  • Emits confidence ratings for every extracted field
POST /v1/document/intelligencecURL
curl -X POST https://api.brainiall.com/v1/document/intelligence \
  -H "Authorization: Bearer $BRAINIALL_KEY" \
  -F "file=@invoice_scan.jpg" \
  -F "extract=key_values,tables"

# Response:
# {
#   "fields": { "invoice_number": "INV-2026-08", "total": "1,420.00", "due_date": "2026-09-15" }
# }

業界別専門スキーマ

財務諸表、医療診断レポート、法的契約書に最適化された事前学習済みモデル。

  • Financial: Invoices, bank statements, balance sheets, tax receipts
  • Healthcare: Patient intake forms, lab results, prescription orders
  • Legal: Deeds, contracts, court dockets, non-disclosure agreements
POST /v1/document/verticals/financialcURL
curl -X POST https://api.brainiall.com/v1/document/verticals/financial \
  -H "Authorization: Bearer $BRAINIALL_KEY" \
  -F "file=@bank_statement.pdf" \
  -F "schema=bank_statement_v2"

大規模バッチ処理パイプライン

1万ページ以上の文書アーカイブを非同期Webhook通知付きで並列高速処理。

  • High-throughput ingestion for archives of 10,000+ pages
  • Webhook completion callbacks with automatic retries on corrupt pages
POST /v1/document/batchcURL
curl -X POST https://api.brainiall.com/v1/document/batch \
  -H "Authorization: Bearer $BRAINIALL_KEY" \
  -d '{
    "document_urls": ["https://s3.../doc1.pdf", "https://s3.../doc2.pdf"],
    "webhook_url": "https://myapp.com/webhooks/docs"
  }'

文書処理を大規模に自動化

エンタープライズスループットで複雑な文書からクリーンなデータを抽出。

無料トライアルを開始
ドキュメント解析・レイアウトOCR API | Brainiall | Brainiall