Snaps

Snaps are the primary mechanism for sending data from your devices to Hub, whether you're growing datasets, capturing field data, or centralizing project information. Snaps are more than just captured data, they are the foundation for building smarter, continuously improving systems. By capturing what your cameras see, Snaps enable you to:

Record and analyze events detected by your devices in the field.
Feed real-world data into analytics pipelines for pattern recognition and performance validation.
Collect and label images or videos to continuously improve the AI models running in your applications.

Snaps make it easy to close the loop between data capture, analysis, and model improvement, helping you create adaptive, reliable, and intelligent vision solutions.

Each Snap represents a discrete event or capture (for example, "image", "data_collection", or "inspection_event"), and you'll typically reuse the same Snap name across many captures.
A Snap can include one or more attachments, such as still images, video sequences, or stereo data.
To make Snap management at scale easier, each Snap supports tags and extras (simple key:value pairs that help organize, search, and filter your captured data).

How to Create Snaps

There are several ways to create and send Snaps, depending on your workflow and level of control:

Data Gathering App

The easiest way to get started — no coding required. Install from the Hub App Store, start capturing from your OAK device, and your Snaps will automatically appear in Luxonis Hub. Useful for initial data collection.

Snaps Producer (DepthAI Integration)

Ideal for building basic custom apps. Use Snaps Producer directly from DepthAI nodes to emit Snaps programmatically.

Direct DAI Integration

For full control and advanced customization. Integrate Snap creation directly into your application logic using the DepthAI API.

Organizing and Searching Snaps

Each Snap can include optional metadata for flexible data management:

Tags, used for grouping related Snaps (e.g., night, validation, dataset_v2).
Extras, arbitrary key-value pairs for searchable metadata (e.g., scene:warehouse, lighting:low).

You can use these fields in Luxonis Hub to search, filter, and organize your captured data at scale.