Custom Properties
Custom properties let you store your own business data against catalogue items, alongside the core and recognition properties that ZIA uses for image recognition.
They exist so that your catalogue in ZIA can carry the same structure you use internally — categories, hierarchies, manufacturer details, internal codes — without forcing that data into fields designed for recognition.
What they are used for
Custom properties are for catalogue management, filtering and export. Typical uses include:
- Grouping items by your own category or hierarchy
- Storing internal references that don't fit the standard Custom ID field
- Holding manufacturer or supplier detail
- Filtering large catalogues down to a working subset
- Carrying your own fields through into exports and reporting
Custom properties do not influence image recognition. The models use the recognition properties (brand, flavour, size, container type, packaging size) and your reference images. Adding a custom property will not change detection or classification accuracy.
Types of custom property
Simple properties
A single value attached to an item. Each definition has a type:
| Type | Example use |
|---|---|
| Text | Internal category name, supplier reference |
| Number | Weight, case quantity, shelf-life in days |
| Boolean | Core range yes/no, promotional item yes/no |
| Date | Launch date, delisting date |
Property groups
When the data you want to store is itself an object rather than a single value, a property group holds several related fields together. For example, a Manufacturer group might carry country, address and company size, and then be assigned to any item that comes from that manufacturer.
Property groups are how catalogue hierarchies are modelled — you define the entity once and reference it from many items, rather than repeating the same values on every item.
How they are set up
Custom property definitions are created by Neurolabs to match your catalogue management needs. Once a definition exists, you can set and edit its values against your items in the web app.
To add, rename or remove a custom property definition, contact your Neurolabs point of contact with the property name, its type, and how you intend to use it. Definitions are organisation-wide, so it's worth agreeing the set up once rather than adding them piecemeal.
Deleting a definition removes the values stored against it for every item, so removals are treated as a deliberate change rather than a routine edit.
Custom properties in results
Custom property values are included in the JSON output for image recognition results, so downstream systems can read your own fields directly from the response instead of joining back to another source.
Direct API endpoints for discovering and managing property definitions (listing definitions, listing the distinct values for a definition) are on the roadmap but are not yet available. Today, definitions are managed by Neurolabs and values are read from the result payload or from an export. If programmatic access to definitions would unblock an integration, raise it with your Neurolabs contact so it can be weighted against other work.
Related
- Important Concepts — core and recognition properties
- Catalogue Management (Quick Start) — day-to-day item and variation management