Caption & Human Re-identification

Please contact us for this API access!

Memories.ai provides advanced APIs to understand media content. With these APIs, you can generate captions and summaries for videos and images, while optionally identifying and tracking specific people using human re-identification (ReID).

Once submitted, your media is analyzed asynchronously, and the results (captions, summaries, or detected individuals) are sent back to your application through a callback URL.

Key Capabilities

• Video Caption
  Automatically summarize video content and describe scenes. Supports human re-identification to detect people based on reference images.
  → Video Caption API

• Image Caption
  Generate captions or descriptions from images. Works with both image URLs and file uploads.
  → Image Caption API

• Human Re-identification (ReID)
  Identify and track specific people in video or image content by providing reference images. This feature is integrated into both Video and Image Caption APIs.
  → Human ReID

Callback Support

You can include a callback URL in your requests to receive results automatically when analysis is complete. This removes the need to poll for updates.

When to Use

• Content Understanding: Summarize long videos or describe images in plain language.
• Security & Monitoring: Detect and re-identify individuals across multiple videos or images.
• Accessibility: Provide captions for visually impaired users or enhance media metadata.

FAQ

• Is Human ReID a separate API?
  No. Human ReID is enabled by adding the persons parameter in Video or Image Caption requests.

• Do I need a callback URL?
  Yes, results are delivered asynchronously via the callback parameter. You must provide a reachable endpoint.

• What are the media limitations?

  • Videos: Max size 20 MB, duration 20–300 seconds
  • Up to 5 person reference images can be included for ReID