Smart Tag template-Intelligent Media Services(IMS)-阿里云帮助中心

This topic describes the parameters for Smart Tag templates and provides corresponding sample templates.

Name	Type	Required	Description	Example
TemplateName	string	Yes	The name of the custom template. The name can contain Chinese and English characters and can be up to 120 characters long.	template-example-****
Industry	string	Yes	The industry of the file to be analyzed. Select an appropriate industry. The system uses the industry to find a suitable algorithm combination. Default value: common (general industry). Valid values: microVideo: The short video industry. common: The general industry.	common
Scene	string	Yes	The use scenario of the file to be analyzed. Select an appropriate scenario. The system uses the scenario to find a suitable algorithm combination. Valid values: search: The search scenario. recommend: The recommendation scenario. Default value: search.	search
AnalyseTypes	string	Yes	The analysis types. You can select multiple types and separate them with commas (,). The following types are supported: ocr: Detects tags from text in images. asr: Detects tags from speech in audio. nlp: Natural Language Processing. classification: Video classification. shows: Detects programs in videos. face: Detects faces in videos or images. role: Detects character roles in videos. object: Detects objects in videos. tvstation: Detects TV station logos in videos. action: Detects action events in videos. emotion: Detects emotions in videos. landmark: Detects landmarks in videos. scene: Detects scenarios in videos. movieIp: Detects movie and TV intellectual property (IP). This feature is available only in the China (Beijing) region. subtitle: The caption extraction feature. Note Make sure that the file you want to analyze contains results for the selected analysis types. Otherwise, the query result may be empty. For example, if you select OCR as the analysis type but the uploaded image does not contain OCR results, an empty result is returned. Billing The mapping between analysis types and billable items is as follows: Video classification and structured tags: Analysis types other than face, ocr, and asr fall under this billable item. The fees are not cumulative. Video facial recognition: Enable face. Video OCR tags: Enable ocr. Video speech recognition tags: Enable asr. Note You are charged for the analysis types that you enable for the DataQ - Smart Tag Service feature. For more information about billing, see DataQ - Smart Tag Service.	ocr
FaceCategoryIds	string	No	The IDs of the person libraries to use for recognition. You can specify multiple IDs. Separate them with commas (,). You can enter the IDs of system person libraries and custom person libraries. The IDs of the system person libraries are as follows: celebrity: Public figures. politician: Political figures. sensitive: Sensitive figures. For information about the IDs of custom person libraries, see Register a custom face.	celebrity
FaceCustomParamsConfig	string	No	The parameters for the facial recognition algorithm. The value must be a JSON string. You can adjust the thresholds for face detection and facial recognition. faceDetThreshold: The threshold for face detection. The default value is 0.999. This parameter is valid only for the analysis of unrecognized faces. faceRegThreshold: The threshold for facial recognition. The default value is 0.9.	{ "faceDetThreshold":0.999, "faceRegThreshold":0.9 }
ObjectGroupIds	string	No	The IDs of the object detection libraries to use. You can specify multiple IDs. Separate them with commas (,). You can enter the IDs of system object detection libraries and custom object detection libraries. The IDs of the system object detection libraries are as follows: general: General. item: Commodity. weapon: Weapon. animal: Animal. For information about the IDs of custom object detection libraries, see the CreateCustomEntity API reference.	general,item,weapon,animal
LandmarkGroupIds	string	No	The IDs of the landmark libraries to use. You can specify multiple IDs. Separate them with commas (,). You can enter the IDs of system landmark libraries and custom landmark libraries. The IDs of the system landmark libraries are as follows: common: General. For more information about the custom region library ID, see the Register Custom Entity Photo API documentation.	common
KeywordConfig	string	No	The configuration for keyword tags. type specifies the part-of-speech categories. You can select multiple categories. Separate them with commas (,). The following categories are supported: name: Person name location: Location name organization: Organization name other: Other Note If you do not specify KeywordConfig, specify an invalid value for KeywordConfig that is not a JSON string, do not include the type field in KeywordConfig, or specify an invalid value for the type field, keywords of all types are returned by default.	"type": "name,location,organization,other" }
LabelCustomCategoryIds	string	No	The IDs of the custom tag libraries to use. You can specify multiple IDs. Separate them with commas (,).	ab*cd,ef*hi
LabelCustomParamsConfig	string	No	The parameter settings for additional custom tag libraries. Only the threshold can be set. Example: {"threshold": xxx}	{"threshold": 0.4}

Sample template

{
	"KeywordConfig": "{\"type\":\"name,location,organization,other\"}",
	"FaceCategoryIds": "celebrity,politician,sensitive",
	"FaceCustomParamsConfig": "{\"faceDetThreshold\":0.9,\"faceRegThreshold\":0.9}",
	"ObjectGroupIds": "general,item,weapon,animal",
	"LandmarkGroupIds": "common",
	"LogoGroupIds": "common",
	"AnalyseTypes": "ocr,asr,classification,shows,face,role,object,tvstation,action,emotion,landmark,scene,organization,keywords,nlp,clipSplit",
	"TemplateName": "newTemplate"
}