Extract information from one or more documents.

Given a document, this endpoint will extract information from it.

Supported document formats are:
- pdf
- jpeg / jpg
- png
- tiff
- webp

Path Parameters

• organization_id integer required

  unique ID of your workspace

• project_id integer required

  is a unique ID of your project

• model_id integer required

  is a unique ID of your trained AI model

Query Parameters

• return_bboxes string

  whether or not to return bounding boxes

• return_annotated_pages string

  whether or not to return the images correspondent to each page, with the found matches, as a base64 encoded string

multipart/form-data

Request Body required

• file binary required

  file to upload

Responses

• 200
• 400
• 401
• 404
• 500

---

OK

application/json

• Schema
• Example (from schema)

---

Schema

• model string

  Name of the model used to extract the information

• all_required_fields_found boolean

  If all required labels were found while processing the documents

• all_confidence_thresholds_met boolean

  If all the expected confidence for the labels were met while processing the documents

• all_data_conversion_passed boolean

  If all data was converted according to the labels configuration while processing the documents

• total_credits_used integer

  Total of credits consumed by the extraction

• documents object[]

  Information extracted from the documents, per document

  name string

  Name of the document

  all_required_fields_found boolean

  If all required labels were found while processing this specific document

  all_confidence_thresholds_met boolean

  If all the expected confidence for the labels were met while processing this specific document

  all_data_conversion_passed boolean

  If all data was converted according to the labels configuration while processing this specific document

  credits_used integer

  Total of credits consumed by the extraction for this specific document

  matches object[]

  All the information extracted, per document

  bboxes array[]

  All the bounding boxes found, per document

  annotated_pages string[]

  All the document pages with annotated date

  hitl object

  Information about the HITL

  review_url string

  the API endpoint to retrieve that review

  status string

  the HITL status (pending or completed)

  result object[]

  All the issues found by the HITL

  confidences object[]

  List all the matches bellow the expected threshold

  label string

  Label associated with the value

  value string

  Value found

  expected_confidence number

  Expected confidence for the label

  found number

  Found confidence for the label

• labels object[]

  All project labels and configuration for each label

  id integer

  The ID of the label

  name string

  The label's name.

{
  "model": "Invoice Extraction v1",
  "all_required_fields_found": true,
  "all_confidence_thresholds_met": true,
  "all_data_conversion_passed": true,
  "total_credits_used": 24,
  "documents": [
    {
      "name": "Invoice.pdf",
      "all_required_fields_found": true,
      "all_confidence_thresholds_met": true,
      "all_data_conversion_passed": true,
      "credits_used": 24,
      "matches": [
        {}
      ],
      "bboxes": [
        [
          [
            {
              "TLx": 0.05868118572292801,
              "TLy": 0.03465982028241335,
              "TRx": 0.06412583182093164,
              "TRy": 0.03465982028241335,
              "BRx": 0.06412583182093164,
              "BRy": 0.04150620453572957,
              "BLx": 0.05868118572292801,
              "BLy": 0.05868118572292801,
              "text": "John Doe",
              "label": "First name",
              "confidence": 0.9936525225639343
            }
          ]
        ]
      ],
      "annotated_pages": [
        "data:image/png;base64,iVBORw0KGgo..."
      ],
      "hitl": {
        "review_url": "https://api.deepopinion.ai/organizations/1/projects/1/hitl/1/reviews/1",
        "status": "pending",
        "result": [
          {
            "confidences": [
              {
                "label": "First name",
                "value": "John Doe",
                "expected_confidence": 90,
                "found": 70
              }
            ]
          }
        ]
      }
    }
  ],
  "labels": [
    {
      "id": 1,
      "name": "POS"
    }
  ]
}

No document was provided. At least one is necessary

User is not authenticated. This means token is either not present or invalid

The project, or model doesn't exist or the user doesn't have the required permission to access the resource.

Something is wrong with the API.

Loading...