DatasetConfig

更新时间:
复制 MD 格式

DatasetConfig is the recommended entry point for configuring dataset-level features, such as content-aware settings and smart grouping. WorkflowParameters is retained only for backward compatibility as a basic workflow switch. New features will no longer be introduced through WorkflowParameters. Configuration options related to reverse image search and person reference are reserved for future use and are not yet available.

Entry points

  • Configure DatasetConfig in OpenMetaQuery to set the bucket-level default processing configuration.

  • Configure datasetConfig in a CreateDataset or UpdateDataset operation to persist an independent configuration for a specific dataset.

  • When RouteRule.AutoCreateDataset=True, datasets that are automatically created by the system may use the default configuration from OpenMetaQuery at runtime. However, this inherited configuration is not guaranteed to be persisted in the dataset record. As a result, a GetDataset call is not guaranteed to return the inherited configuration. For configurations that must be explicitly queried and maintained, you must set them in datasetConfig when calling CreateDataset or UpdateDataset.

  • If you explicitly call CreateDataset and do not pass the datasetConfig parameter, the DatasetConfig from OpenMetaQuery is not copied to the dataset's persistent configuration.

Parameter

Type

Required

Example

Description

DatasetConfig

container

No

N/A

A container for dataset-level configurations.

Child nodes: Insights, ReverseImage, SmartCluster

Insights

container

No

N/A

A container for detailed content-aware settings.

Child nodes: Language, Image, Video

Parent node: DatasetConfig

Language

string

No

zh-Hans

The language for text generated by Insights. The default value is zh-Hans. Supported language codes: zh-Hans (Simplified Chinese), zh-Hant (Traditional Chinese), en (English), ja (Japanese), ko (Korean), ar (Arabic), cs (Czech), da (Danish), nl (Dutch), fr (French), de (German), id (Indonesian), it (Italian), ms (Malay), no (Norwegian), pl (Polish), pt (Portuguese), ro (Romanian), ru (Russian), es (Spanish), sv (Swedish), th (Thai), and tr (Turkish).

This setting can be overridden for an individual file by using the Object Storage Service (OSS) object tag x-oss-metaquery-insights-language.

Parent node: Insights

Image

container

No

N/A

A container for image-specific content-aware settings.

Child node: Caption

Parent node: Insights

Caption

container

No

N/A

A container for image caption settings.

Child nodes: Enable, Prompt

Parent node: Image (under Insights)

Enable

boolean

No

True

Enables caption generation for images. The default value is False.

Parent node: Caption (under Insights.Image)

Prompt

string

No

Describe the scene in one sentence.

A custom prompt for image captioning.

Parent node: Caption (under Insights.Image)

Video

container

No

N/A

A container for video-specific content-aware settings.

Child nodes: Caption, Label, MultiStream

Parent node: Insights

Caption

container

No

N/A

A container for video caption settings.

Child nodes: Enable, Prompt, PersonReference

Parent node: Video (under Insights)

Enable

boolean

No

True

Enables caption generation for videos. The default value is False.

Parent node: Caption (under Insights.Video)

Prompt

string

No

Summarize the video in one sentence.

A custom prompt for video captioning.

Parent node: Caption (under Insights.Video)

PersonReference

container

No

N/A

A reserved container for person reference settings in video captions. This parameter is for future use and has no effect if configured.

Child node: Enable

Parent node: Caption (under Insights.Video)

Enable

boolean

No

N/A

A reserved switch to enable person reference in video captions. This parameter is for future use and has no effect if configured. The default value is False.

Parent node: PersonReference

Label

container

No

N/A

A container for video label settings.

Child node: UserDefined

Parent node: Video (under Insights)

UserDefined

container

No

N/A

A container for user-defined video label settings.

Child nodes: Enable, Labels

Parent node: Label

Enable

boolean

No

True

Enables user-defined labels. The default value is False.

Parent node: UserDefined

Labels

container

No

N/A

A container for the list of user-defined labels.

Child node: Label

Parent node: UserDefined

Label

container

No

N/A

A container for a single user-defined label.

Child nodes: Name, Description

Parent node: Labels

Name

string

Yes

A person has fallen

The name of the user-defined label.

Parent node: Label (under UserDefined.Labels)

Description

string

No

The action of a person in the frame changing from a standing or walking posture to a fallen position on the ground.

The description of the user-defined label. This helps the model understand the trigger conditions.

Parent node: Label (under UserDefined.Labels)

MultiStream

container

No

N/A

A container for parallel multi-stream video analysis settings.

Child node: Enable

Parent node: Video (under Insights)

Enable

boolean

No

True

Enables parallel multi-stream video analysis. The default value is False.

Parent node: MultiStream

ReverseImage

container

No

N/A

A reserved container for reverse image search settings. This parameter is for future use and has no effect if configured.

Child nodes: Image, Video

Parent node: DatasetConfig

Image

container

No

N/A

A reserved container for image-based reverse image search settings. This feature is reserved for future use.

Child node: Enable

Parent node: ReverseImage

Enable

boolean

No

N/A

A reserved switch to enable image-based reverse image search. This parameter is for future use and has no effect if configured. The default value is False.

Parent node: Image (under ReverseImage)

Video

container

No

N/A

A reserved container for video-based reverse image search settings. This feature is reserved for future use.

Child node: Enable

Parent node: ReverseImage

Enable

boolean

No

N/A

A reserved switch to enable video-based reverse image search. This parameter is for future use and has no effect if configured. The default value is False.

Parent node: Video (under ReverseImage)

SmartCluster

container

No

N/A

A container for dataset-level automatic smart grouping settings.

Child node: Figure

Parent node: DatasetConfig

Figure

container

No

N/A

A container for face clustering settings.

Child nodes: AutoGenerate, AutoClustering, MinEntityCount, EnabledFeatures

Parent node: SmartCluster

AutoGenerate

boolean

No

True

Enables automatic generation of face clusters. When enabled, faces are automatically clustered as they are added to the library. The default value is False.

Parent node: Figure

AutoClustering

boolean

No

True

Enables continuous merging and splitting of existing clusters. The default value is False.

Parent node: Figure

MinEntityCount

integer

No

3

The minimum number of faces required to form a cluster. The default value is 3.

Parent node: Figure

EnabledFeatures

list of strings

No

face

The additional features to enable. Currently, only face is supported. To specify multiple values, include this element multiple times.

Parent node: Figure