Video AI parameter descriptions

更新时间:
复制 MD 格式

This topic describes the parameters for the automated review, smart thumbnail, media fingerprint, and DataQ - Smart Tag Service APIs of ApsaraVideo VOD.

AI template content information (AITemplateInfo)

Name

Type

Required

Description

TemplateId

String

Yes

The template ID.

TemplateType

String

Yes

The template type. Valid values:

  • AIMediaAudit: The automated review template.

  • AIImage: The smart thumbnail template.

TemplateName

String

Yes

The template name.

TemplateConfig

String

Yes

The detailed template configuration. This is a JSON string. For more information, see AITemplateConfig.

Source

String

Yes

The source of the template. Valid values:

  • System: system

  • Custom: A user-defined option.

IsDefault

String

No

Indicates whether the template is the default one. Valid values:

  • Default: Yes.

  • NotDefault: No.

CreationTime

String

Yes

The time when the template was created. The time is in UTC format.

ModifyTime

String

Yes

The time of the last modification in UTC format.

AI template configuration (AITemplateConfig)

When TemplateType is AIMediaAudit

Field

Type

Required

Description

AuditItem

String[]

Yes

The automated review items. Valid values:

  • terrorism: Terrorism-related content.

  • porn: Pornography detection.

  • ad: Advertisements.

  • Logo: icon

  • live: Undesirable scenes.

  • audio: Audio anti-spam.

Note
  • If the input file for review has an audio track and you set this parameter to audio, the audio is reviewed. If you do not set this parameter to audio, the audio is not reviewed.

  • If the input file is audio-only, you must set this parameter to audio. If the input file does not have an audio track, do not set this parameter to audio.

AuditRange

String[]

Yes

The scope of the automated review. Valid values:

  • video: The video.

  • image-cover: The thumbnail.

  • text-title: The title.

AuditContent

String[]

Yes

The content for automated review. The value is screen.

AuditAutoBlock

String

Yes

Indicates whether to automatically block the video if it is flagged for violations. Valid values:

  • no: No.

  • yes: Yes.

When TemplateType is AIImage

Field name

Type

Required

Description

Format

String

Yes

The format of the smart thumbnail. Valid values:

  • gif

  • png

SetDefaultCover

String

Yes

Indicates whether to set the generated image as the video thumbnail. Valid values:

  • true: After the job is successful, an image is selected and set as the video thumbnail.

  • false: No thumbnail is set for the video.

Automated review result (AIMediaAuditResult)

Field name

Type

Required

Description

Suggestion

String

Yes

The suggested action for the review result. Valid values:

  • Blocked due to a violation.

  • Review result: Suspected

  • Pass: Passed.

AbnormalModules

String

No

The content that is flagged for violations. Multiple values are separated by commas (,). Valid values:

  • video: The video.

  • image-cover: The thumbnail.

  • text-title: The title.

Label

String

Yes

The category of the review result. Multiple values are separated by commas (,). Valid values:

  • porn: Pornography.

  • terrorism: Terrorism and political content.

  • ad: Text and image violations.

  • live: Undesirable scenes.

  • logo: Image logos.

  • audio: Audio anti-spam.

  • normal: Normal.

VideoResult

VideoResult

No

The video review result.

CoverResult

Note

In the response to a GetMediaAuditResult request, this field is named ImageResult.

CoverResult[]

No

The video thumbnail review result.

TitleResult

Note

In the response to a GetMediaAuditResult request, this field is named TextResult.

TitleResult[]

No

The video title review result.

AudioResult

AudioResult[]

No

The audio review result.

ImageResult

Note

This field is returned only in the callback message for a SubmitAIMediaAuditJob request for image review.

ImageResult[]

No

The image review result.

VideoResult

Field name

Type

Required

Description

Suggestion

String

Yes

The suggested action for the review result. Valid values:

  • Blocked due to a violation.

  • Review result: Suspected.

  • pass: The check passed.

Label

String

Yes

The category of the review result. Multiple values are separated by commas (,). Valid values:

  • porn: Pornography.

  • terrorism: Terrorism and political content.

  • ad: Text and image violations.

  • live: Undesirable scenes.

  • logo: Image logos.

  • normal: Normal.

TerrorismResult

TerrorismResult[]

No

The review result for terrorism and political content.

PornResult

PornResult[]

No

The review result for pornography.

AdResult

AdResult[]

No

The review result for advertisements.

LiveResult

LiveResult[]

No

The review result for undesirable content.

LogoResult

LogoResult[]

No

The review result for logos.

TerrorismResult

Field name

Type

Required

Value

Suggestion

String

Yes

The suggested action for the review of terrorism and political content. Valid values:

  • Blocked: Violation

  • Review status: Suspected.

  • pass: Approve the content.

Label

String

Yes

The review result for terrorism and political content. Valid values:

  • normal: Normal.

  • bloody: Bloody.

  • explosion: Explosions and smoke.

  • outfit: Special attire.

  • A logo is a unique identifier.

  • weapon: Weapons.

  • politics: Political content.

  • violence: Fighting.

  • Crowd: Gathering.

  • A parade is a type of procession.

  • carcrash: Car accidents.

  • flag: Flags.

  • location: Landmarks.

  • others: Others.

MaxScore

String

Yes

The highest score of a video snapshot that hits the label. The value ranges from 0 to 100. The value is accurate to 10 decimal places.

A higher score indicates a higher probability that the content belongs to the label.

AverageScore

String

Yes

The average score of the video snapshots that hit the label. The value ranges from 0 to 100. The value is accurate to 10 decimal places.

A higher score indicates a higher probability that the content belongs to the label.

CounterList

TerrorismCounterList[]

Yes

The number of video snapshots for each category of terrorism and political content.

TopList

TerrorismTopList[]

Yes

The information about the top 10 video snapshots with the highest scores for the label.

TerrorismCounterList

Field name

Type

Required

Value

Label

String

Yes

The review result for terrorism and political content. Valid values:

  • normal: Normal.

  • bloody: Bloody.

  • explosion: Explosions and smoke.

  • outfit: Special attire.

  • logo: A unique identifier.

  • Weapon: weapon.

  • politics: Political content.

  • violence: Fighting.

  • A crowd is a gathering of people.

  • parade: Parades.

  • carcrash: Car accidents.

  • flag: Flags.

  • location: Landmarks.

  • others: Others.

Count

Integer

Yes

The number of video snapshots.

TerrorismTopList

Field name

Type

Required

Value

Label

String

Yes

The review result for terrorism and political content. Valid values:

  • normal: Normal.

  • Bloody means gory.

  • explosion: Explosions and smoke.

  • outfit: Special attire.

  • logo: A unique identifier.

  • weapon: Weapons.

  • politics: Political content.

  • violence: Fighting.

  • crowd: Crowds.

  • parade: Parades.

  • carcrash: A scene of a car crash.

  • flag: Flags.

  • location: Landmarks.

  • others: Others.

Score

String

Yes

The score of the video snapshot that hits the label. The value ranges from 0 to 100. The value is accurate to 10 decimal places.

A higher score indicates a higher probability that the content belongs to the label.

Timestamp

String

Yes

The position of the video snapshot in the video. Unit: milliseconds.

Url

String

Yes

The URL of the video snapshot.

PornResult

Field name

Type

Required

Value

Suggestion

String

Yes

The suggested action for the pornography review. Valid values:

  • Block for violation.

  • Review result: Suspected

  • Pass: Passed.

Label

String

Yes

The pornography review result. Valid values:

  • porn: Pornography.

  • sexy: sexy.

  • Normal.

MaxScore

String

Yes

The highest score of a video snapshot that hits the label. The value ranges from 0 to 100. The value is accurate to 10 decimal places.

A higher score indicates a higher probability that the content belongs to the label.

AverageScore

String

Yes

The average score of the video snapshots that hit the label. The value ranges from 0 to 100. The value is accurate to 10 decimal places.

A higher score indicates a higher probability that the content belongs to the label.

CounterList

PornCounterList[]

Yes

The number of video snapshots for each category of the review result.

TopList

PornTopList[]

Yes

The information about the top 10 video snapshots with the highest scores for the label.

PornCounterList

Field

Type

Required

Value

Label

String

Yes

The pornography review result. Valid values:

  • porn: Pornography.

  • sexy: Sexy.

  • normal: Normal.

Count

Integer

Yes

The number of video snapshots.

PornTopList

Field name

Type

Required

Value

Label

String

Yes

The pornography review result. Valid values:

  • porn: Pornography.

  • sexy: Sexy.

  • Normal.

Score

String

Yes

The score of the video snapshot that hits the label. The value ranges from 0 to 100. The value is accurate to 10 decimal places.

A higher score indicates a higher probability that the content belongs to the label.

Timestamp

String

Yes

The position of the video snapshot in the video. Unit: milliseconds.

Url

String

Yes

The URL of the video snapshot.

AdResult

Field name

Type

Required

Value

Suggestion

String

Yes

The suggested action for the advertisement review. Valid values:

  • Blocked: Violation.

  • Review result: Suspected

  • pass: The operation was successful.

Label

String

Yes

The advertisement review result. Valid values:

  • normal: Normal.

  • ad: Other advertisements.

  • politics: Text that contains political content.

  • porn: Text that contains pornographic content.

  • abuse: Text that contains abusive content.

  • terrorism: Text that contains terrorism-related content.

  • contraband: Text that contains prohibited content.

  • spam: Text that contains other spam.

  • npx: Undesirable ads.

  • qrcode: QR codes.

  • programCode: Mini Program codes.

MaxScore

String

Yes

The highest score of a video snapshot that hits the label. The value ranges from 0 to 100. The value is accurate to 10 decimal places.

A higher score indicates a higher probability that the content belongs to the label.

AverageScore

String

Yes

The average score of the video snapshots that hit the label. The value ranges from 0 to 100. The value is accurate to 10 decimal places.

A higher score indicates a higher probability that the content belongs to the label.

CounterList

AdCounterList[]

Yes

The number of video snapshots for each category of the advertisement review result.

TopList

AdTopList[]

Yes

The information about the top 10 video snapshots with the highest scores for the label.

AdCounterList

Field

Type

Required

Value

Label

String

Yes

The category of the advertisement review result. Valid values:

  • normal: Normal.

  • ad: Other advertisements.

  • politics: Text that contains political content.

  • porn: Text that contains pornographic content.

  • abuse: Text that contains abusive content.

  • terrorism: Text that contains terrorism-related content.

  • contraband: Text that contains prohibited content.

  • spam: Text that contains other spam.

  • npx: Undesirable ads.

  • qrcode: QR codes.

  • programCode: Mini Program codes.

Count

Integer

Yes

The number of video snapshots.

AdTopList

Field Name

Type

Required

Value

Label

String

Yes

The category of the advertisement review result. Valid values:

  • normal: Normal.

  • ad: Other advertisements.

  • politics: Text that contains political content.

  • porn: Text that contains pornographic content.

  • abuse: Text that contains abusive content.

  • terrorism: Text that contains terrorism-related content.

  • contraband: Text that contains prohibited content.

  • spam: Text that contains other spam.

  • npx: Undesirable ads.

  • qrcode: QR codes.

  • programCode: Mini Program codes.

Score

String

Yes

The score of the video snapshot that hits the label. The value ranges from 0 to 100. The value is accurate to 10 decimal places.

A higher score indicates a higher probability that the content belongs to the label.

Timestamp

String

Yes

The position of the video snapshot in the video. Unit: milliseconds.

Url

String

Yes

The URL of the video snapshot.

LiveResult

Field name

Type

Required

Value

Suggestion

String

Yes

The suggested action for the review of undesirable content. Valid values:

  • Blocked due to violation.

  • Review Result: Suspected.

  • pass: Approve the content.

Label

String

Yes

The review result for undesirable content. Valid values:

  • normal: Normal.

  • meaningless: The image has no content, such as a black screen or a white screen.

  • PIP: Picture-in-Picture (PiP).

  • smoking: Smoking.

  • drivelive: Live streaming from inside a vehicle.

MaxScore

String

Yes

The highest score of a video snapshot that hits the label. The value ranges from 0 to 100. The value is accurate to 10 decimal places.

A higher score indicates a higher probability that the content belongs to the label.

AverageScore

String

Yes

The average score of the video snapshots that hit the label. The value ranges from 0 to 100. The value is accurate to 10 decimal places.

A higher score indicates a higher probability that the content belongs to the label.

CounterList

LiveCounterList[]

Yes

The number of video snapshots for each category of the undesirable content review result.

TopList

LiveTopList[]

Yes

The information about the top 10 video snapshots with the highest scores for the label.

LiveCounterList

Field

Type

Required

Value

Label

String

Yes

The category of the undesirable content review result. Valid values:

  • normal: Normal.

  • meaningless: The image has no content, such as a black screen or a white screen.

  • PIP: Picture-in-Picture (PiP).

  • smoking: Smoking.

  • drivelive: Live streaming from inside a vehicle.

Count

Integer

Yes

The number of video snapshots.

LiveTopList

Field

Type

Required

Value

Label

String

Yes

The category of the undesirable content review result. Valid values:

  • normal: Normal.

  • meaningless: The image has no content, such as a black screen or a white screen.

  • PIP: Picture-in-Picture (PiP).

  • smoking: Smoking.

  • drivelive: Live streaming from inside a vehicle.

Score

String

Yes

The score of the video snapshot that hits the label. The value ranges from 0 to 100. The value is accurate to 10 decimal places.

A higher score indicates a higher probability that the content belongs to the label.

Timestamp

String

Yes

The position of the video snapshot in the video. Unit: milliseconds.

Url

String

Yes

The URL of the video snapshot.

LogoResult

Field name

Type

Required

Value

Suggestion

String

Yes

The suggested action for the logo review. Valid values:

  • Block violation.

  • Review result: Suspected

  • Pass: Passed

Label

String

Yes

The review result for logo content. Valid values:

  • normal: Normal.

  • TV: Contains controlled logos.

  • trademark: Contains trademarks.

MaxScore

String

Yes

The highest score of a video snapshot that hits the label. The value ranges from 0 to 100. The value is accurate to 10 decimal places.

A higher score indicates a higher probability that the content belongs to the label.

AverageScore

String

Yes

The average score of the video snapshots that hit the label. The value ranges from 0 to 100. The value is accurate to 10 decimal places.

A higher score indicates a higher probability that the content belongs to the label.

CounterList

LogoCounterList[]

Yes

The number of video snapshots for each category of the logo review result.

TopList

LogoTopList[]

Yes

The information about the top 10 video snapshots with the highest scores for the label.

LogoCounterList

Field

Type

Required

Value

Label

String

Yes

The category of the logo review result. Valid values:

  • normal: Normal.

  • TV: Contains controlled logos.

  • trademark: Contains trademarks.

Count

Integer

Yes

The number of video snapshots.

LogoTopList

Field name

Type

Required

Value

Label

String

Yes

The category of the logo review result. Valid values:

  • normal: Normal.

  • TV: Contains controlled logos.

  • trademark: Contains trademarks.

Score

String

Yes

The score of the video snapshot that hits the label. The value ranges from 0 to 100. The value is accurate to 10 decimal places.

A higher score indicates a higher probability that the content belongs to the label.

Timestamp

String

Yes

The position of the video snapshot in the video. Unit: milliseconds.

Url

String

Yes

The URL of the video snapshot.

CoverResult

Field name

Type

Required

Description

Suggestion

String

Yes

The suggested action for the video thumbnail review. Valid values:

  • Blocked due to violation.

  • Review result: Suspicious.

  • pass: Approve the content.

Label

String

Yes

The category of the video thumbnail review result. Multiple values are separated by commas (,). Valid values:

  • porn: Pornography.

  • terrorism: Terrorism and political content.

  • ad: Text and image violations.

  • live: Undesirable scenes.

  • logo: Image logos.

  • normal: Normal.

Type

String

Yes

The review type. The value is cover.

Url

String

Yes

The URL of the video thumbnail.

Result

CoverResultItem[]

Yes

The details of the video thumbnail review result.

CoverResultItem

Field

Type

Required

Value

Suggestion

String

Yes

The suggested action for the video thumbnail review. Valid values:

  • Blocked due to a violation.

  • Review status: Suspected.

  • pass: Successful.

Label

String

Yes

The category of the video thumbnail review result.

If scene is porn, valid values are:

  • porn: Pornography.

  • sexy: Sexy.

  • normal: Normal.

If scene is terrorism, valid values are:

  • normal: Normal.

  • bloody: Bloody.

  • explosion: Explosions and smoke.

  • outfit: Special attire.

  • The logo is a special identifier.

  • weapon: Weapons.

  • politics: Political content.

  • violence: Fighting.

  • crowd: Crowds.

  • parade: Parades.

  • carcrash: Car accidents.

  • flag: Flags.

  • location: Landmarks.

  • others: Others.

If scene is ad, valid values are:

  • normal: Normal.

  • ad: Other advertisements.

  • politics: Text that contains political content.

  • porn: Text that contains pornographic content.

  • abuse: Text that contains abusive content.

  • terrorism: Text that contains terrorism-related content.

  • contraband: Text that contains prohibited content.

  • spam: Text that contains other spam.

  • npx: Undesirable ads.

  • qrcode: QR codes.

  • programCode: Mini Program codes.

If scene is live, valid values are:

  • normal: Normal.

  • meaningless: The image has no content, such as a black screen or a white screen.

  • PIP: Picture-in-Picture (PiP).

  • smoking: Smoking.

  • drivelive: Live streaming from inside a vehicle.

If scene is logo, valid values are:

  • normal: Normal.

  • TV: Contains controlled logos.

  • trademark: Contains trademarks.

Score

String

Yes

The score of the video thumbnail that hits the label. The value ranges from 0 to 100. The value is accurate to 10 decimal places.

A higher score indicates a higher probability that the content belongs to the label.

Scene

String

Yes

The review scenario for the video thumbnail. Valid values:

  • porn: Pornography detection.

  • terrorism: Terrorism and political content.

  • ad: Text and image violations.

  • live: Undesirable scenes.

  • logo: Image logos.

TitleResult

Field

Type

Required

Value

Suggestion

String

Yes

The suggested action for the text review. Valid values:

  • Block: Violation

  • Review result: Questionable.

  • pass: Passed

Label

String

Yes

The category of the text review result. Valid values:

  • spam: Spam.

  • ad: Advertisements.

  • abuse: Abusive content.

  • flood: Posting many messages in a short period.

  • contraband: Prohibited content.

  • meaningless: Meaningless content.

  • normal: Normal.

Score

String

Yes

The score for a Label hit is a value in the range of 0 to 100, with a precision of 10 decimal places.

A higher score indicates a higher probability that the content belongs to the label.

Scene

String

Yes

The review scenario. The value is antispam.

Type

String

Yes

The text category. The value is title.

AudioResult

Field

Type

Required

Value

Suggestion

String

Yes

The suggested action for the audio review. Valid values:

  • Blocked due to a violation.

  • Review result: Suspected.

  • Pass: Passed

Label

String

Yes

The category of the audio review result. Valid values:

  • normal: Normal.

  • spam: Contains unwanted or unsolicited messages.

  • ad: Advertisements.

  • politics: Political content.

  • terrorism: Terrorism-related content.

  • abuse: Abusive content.

  • porn: Pornography.

  • flood: Flooding.

  • contraband: Prohibited content.

  • Meaningless: meaningless.

Score

String

Yes

The score of the audio that hits the label. The value ranges from 0 to 100. The value is accurate to 10 decimal places.

A higher score indicates a higher probability that the content belongs to the label.

Scene

String

Yes

The review scenario. The value is fixed to antispam.

ImageResult

Field name

Type

Required

Description

Suggestion

String

Yes

The suggested action for the image review. Valid values:

  • Blocked: Violation.

  • Review result: Suspected.

  • Status: Passed.

Label

String

Yes

The category of the image review result. Multiple values are separated by commas (,). Valid values:

  • porn: Pornography.

  • terrorism: Terrorism and political content.

  • ad: Text and image violations.

  • live: Undesirable scenes.

  • logo: Image logos.

  • normal: Normal.

Url

String

Yes

The URL of the image.

Result

ImageResultItem[]

Yes

The details of the image review result.

ImageResultItem

Field

Type

Required

Value

Suggestion

String

Yes

The suggested action for the image review. Valid values:

  • block: A violation was detected.

  • Flagged for review.

  • Pass: Passed

Label

String

Yes

The category of the image review result.

If scene is porn, valid values are:

  • porn: Pornography.

  • sexy: Sexy.

  • normal: Normal.

If scene is terrorism, valid values are:

  • normal: Normal.

  • Bloody means gory.

  • explosion: Explosions and smoke.

  • outfit: Special attire.

  • logo: A specific identifier.

  • weapon: Weapons.

  • politics: Political content.

  • violence: Fighting.

  • crowd: Many people gathered together.

  • parade: Parades.

  • carcrash: Car accidents.

  • flag: Flags.

  • location: Landmarks.

  • others: Others.

If scene is ad, valid values are:

  • normal: Normal.

  • ad: Other advertisements.

  • politics: Text that contains political content.

  • porn: Text that contains pornographic content.

  • abuse: Text that contains abusive content.

  • terrorism: Text that contains terrorism-related content.

  • contraband: Text that contains prohibited content.

  • spam: Text that contains other spam.

  • npx: Undesirable ads.

  • qrcode: QR codes.

  • programCode: Mini Program codes.

If scene is live, valid values are:

  • normal: Normal.

  • meaningless: The image has no content, such as a black screen or a white screen.

  • PIP: Picture-in-Picture (PiP).

  • smoking: Smoking.

  • drivelive: Live streaming from inside a vehicle.

If scene is logo, valid values are:

  • normal: Normal.

  • TV: Contains controlled logos.

  • trademark: Contains trademarks.

Score

String

Yes

The score of the image that hits the label. The value ranges from 0 to 100. The value is accurate to 10 decimal places.

A higher score indicates a higher probability that the content belongs to the label.

Scene

String

Yes

The image review scenario. Valid values:

  • porn: Pornography detection.

  • terrorism: Terrorism and political content.

  • ad: Text and image violations.

  • live: Undesirable scenes.

  • logo: Image logos.

Media fingerprint result (AIMediaDNAResult)

Name

Type

Required

Value

VideoDNA

VideoDNA[]

No

The media fingerprint detection result.

VideoDNA

Name

Type

Required

Value

PrimaryKey

String

Yes

The ID of the similar video.

Similarity

String

Yes

The video similarity.

Detail

VideoDNADetail[]

Yes

The details of the similar video, including the position and duration.

VideoDNADetail

Name

Type

Required

Value

Input

VideoDNADetailItem

Yes

The start time and duration of the input video.

Duplication

VideoDNADetailItem

Yes

The start time and duration of the video in the library.

VideoDNADetailItem

Name

Type

Required

Value

Start

String

Yes

The start time of the video.

Duration

String

Yes

The duration of the video.

Smart tag result (AIVideoTagResult)

Field name

Type

Required

Value

Person

AIVideoTagResultItem[]

No

A collection of person tags.

Time

AIVideoTagResultItem[]

No

A collection of time tags.

Location

AIVideoTagResultItem[]

No

A collection of location tags.

Keyword

AIVideoTagResultItem[]

No

A collection of keyword tags.

Category

AIVideoTagResultItem[]

No

A collection of video categories.

OrigASRData

AIASRData[]

No

The raw automatic speech recognition (ASR) result. This parameter is returned only if the AnalyseTypes parameter in Config > AIVideoTag of a SubmitAIJob request includes ASR, and OrigASRData in Config > AIVideoTag is set to true.

AIVideoTagResultItem

Field name

Type

Required

Value

Tag

String

Yes

The tag.

Times

String[]

No

A collection of time points. Unit: milliseconds.

FaceUrl

String

No

The URL of the face. Note: This parameter is returned only for person tags.

Speech recognition result (AIASRData)

Field name

Type

Required

Value

Duration

String

Yes

The audio duration. Unit: milliseconds.

AsrTextList

AsrText[]

No

The speech recognition result. The elements in the array are sentences. For more information, see AsrText.

Sentence result (AsrText)

Field

Type

Required

Value

StartTime

Long

Yes

The start time offset of the sentence. Unit: milliseconds.

EndTime

Long

Yes

The end time offset of the sentence. Unit: milliseconds.

Text

String

Yes

The recognized text of the sentence.

ChannelId

Int

Yes

The sound channel ID.

SpeechRate

Int

Yes

The average speech rate of the sentence. Unit: words per minute.