Digizuite currently offers support for two services:
Ximilar https://www.ximilar.com/
Fashion & Retail
Inscene AI https://thebdx.com/technology/inscene/
Good for home builders and real estate in US
Imagga AI - https://imagga.com/
Great fit for real estate, color tagging, retail and commerce
Pre-requisites:
You need your own account and subscription at Optidash. From your account, the actual API Key must be sent to Digizuite so we know where the request should be directed to.
How it works:
Digizuite offers an automation which can be triggered based on your requirements. Often it will be on all new product images being uploaded into the platform. Since this is in our automation, it is completely customized what should trigger so you could do it for certain metadata or workflow stages.
Endpoints
These endpoints are the ones which is used in the automation config further down
Type | Endpoint | Input | Functionality | Output |
|---|---|---|---|---|
Ximilar AI |
Since this is custom categories it requires a workshop to define categories and get images which corresponds to those categories for training the AI model. |
| Sends image for tagging at ai service and returns tags. | Returns interface DigizuiteTagResponse {
status: string; // from ai service
errorMessage: string; // from ai service
tags: DetectTagObject[];
}
interface DetectTagObject {
prop: number;
name: number;
id: number;
}
|
Ximiar Fashion Tagging AI |
|
| Sends image for tagging at ai service and returns tags. | Returns interface DigizuiteTagResponse {
status: string; // from ai service
errorMessage: string; // from ai service
tags: DetectTagObject[];
}
interface DetectTagObject {
prop: number;
name: number;
id: number;
}
|
Ximiar Fashion Tagging AI |
|
| Sends image for tagging at ai service and returns tags. | Returns interface DigizuiteTagResponse {
status: string; // from ai service
errorMessage: string; // from ai service
tags: DetectTagObject[];
}
interface DetectTagObject {
prop: number;
name: number;
id: number;
}
|
Inscene AI |
** Be aware here that Inscene AI is only supported in the US. So therefore it has its own URL (dgz-integration-broker-us). |
| interface DigizuiteTagResponse {
status: string; // from ai service
errorMessage: string; // from ai service
tags: DetectTagObject[];
subTags: DetectTagObject[]; // only for Inscene AI
objects: DetectSubObject[]; // only for Inscene AI
}
interface DetectTagObject {
prop: number;
name: number;
id: number;
}
interface DetectSubObject {
prop: number;
name: number;
id: number;
subObjects: DetectTagObject[];
}
| |
Imagga Tagging AI |
|
| Sends image for tagging at ai service and returns tags. | Returns interface DigizuiteTagResponse {
status: string; // from ai service
errorMessage: string; // from ai service
tags: DetectTagObject[];
}
interface DetectTagObject {
prop: number;
name: number;
id: number;
}
|
Configuration
The following shows a starting point for your integration.
Note that client id and secret code must be provided for the endpoint by Digizuite R&D. Also, the ximilar or inscene API key must be provided to Digizuite so it can be associated with the client id.
When you have added the below then you are good to go.
trigger "Metadata trigger" {
type = "Specific Metadata Value"
resolves = "Set String Metafield"
value = "#54a0ff;creative"
ignore_casing = "true"
meta_field = "guid:6fe35f10-c810-497b-af9c-e52c3d583593"
listen_to_metadata_changes = "all"
}
action "Invoke Endpoint" {
type = "Invoke Endpoint"
endpoint = "https://dgz-integration-broker.azurewebsites.net/api/integrations/asset/invoke?clientId={SECRET_CLIENT_ID}&integrationType=ximilaraiservice&assetId=<assetid>&code={SECRET_CODE}"
method = "post"
response = "@response"
}
action "Query JSON" {
type = "Query JSON"
needs = "Invoke Endpoint"
json = "@response"
json_path = "$.result.tags[*].name"
error_on_not_found_items = "true"
result = "@queryResult"
}
action "Set String Metafield" {
type = "Set String Metafield"
needs = "Query JSON"
meta_field = "guid:7441267e-4f7c-4dcd-9c42-20273a4ab6af"
new_value = "@queryResult"
asset_item_ids = "@sourceAssetItemId"
}