Quick Start

Overview

The Vision API enables you to bring human-like reasoning capabilities to your video and imagery data. Upload any format of video from any camera, environment, or industry and leverage spatial and temporal reasoning beyond metadata-based tagging and traditional vision approaches.

Authentication

All API requests require a an API token. Tokens are secrets and should be kept securely and privately. Do not share it with others or expose it in any client-side code. Requests should include your API token in an Authorization HTTP header as follows:

Authorization: Bearer $API_TOKEN

Manage your API token through the access management options (Users).

Basic Workflow

Upload and query an Input.

1. Get a pre-signed URL for video or image upload

curl -X POST "https://api.ondeckai.com/api/v1/pre-signed-url" \
  -H "Authorization: Bearer $API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "file_name": "my-video.mp4"
  }'

Go to endpoint documentation

Response Example

  {
    "input_id": "6c660cec-794e-40a4-a5f2-47f57b1a5b50",
    "signed_url": "https://aaa.bbb.ccc.com/dir/usr/my-video.mp4?X-Amz...",
    "file_name": "my-video.mp4",
    "mime_type": "video/mp4",
    "status": "processing",
    "project_id": "1614184e-84cb-4913-aa32-0b9885d427ec",
    "collection_ids": [],
    "details": {}
  }

2. Upload your video or image file

Use the signed_url from step 1 to upload your file. Once uploaded, inputs are updated with a valid status if it’s format is supported.

  curl -T /path/to/my-video.mp4 "$SIGNED_URL"

3. Query

Query against your video or image Input with free-form natural language.

  curl --request POST \
    --url https://api.ondeckai.com/api/v1/inputs/{id}/queries \
    --header 'Authorization: Bearer <token>' \
    --header 'Content-Type: application/json' \
    --data '{
    "question": "What's happening in this video?",
    "query_type": "synchronous"
  }'

Go to endpoint documentation

Response Example

  {
    "id": "3c90c3cc-0d44-4b50-8888-8dd25736052a",
    "project_id": "1614184e-84cb-4913-aa32-0b9885d427ec",
    "input_id": "6c660cec-794e-40a4-a5f2-47f57b1a5b50",
    "collection_id": null,
    "query_type": "synchronous",
    "status": "completed",
    "question": "What's happening in this video?",
    "short_answer": "The video shows...",
    "answer_uri": null,
    "structured_answer": null
  }

Advanced Workflows

Core Concepts

Understand inputs, collections, inference jobs, queries, and predictions.

Management Features

Set up organizations, users, and projects for granular access control.

Collection Querying

Aggregate your entire dataset in a collection and search at scale.

Configure custom background queries

Define a query that runs automatically and returns exactly what you need with structured responses.

⌘I

Getting started

Core concepts

Management

Status Codes

Overview

Authentication

Basic Workflow

1. Get a pre-signed URL for video or image upload

2. Upload your video or image file

3. Query

Advanced Workflows

Core Concepts

Management Features

Collection Querying

Configure custom background queries

​Overview

​Authentication

​Basic Workflow

1. Get a pre-signed URL for video or image upload

2. Upload your video or image file

3. Query

​Advanced Workflows

Core Concepts

Management Features

Collection Querying

Configure custom background queries

Overview

Authentication

Basic Workflow

Advanced Workflows