All Products
Search

This Product

    :FACE_DEEPFAKE

    Document Center

    :FACE_DEEPFAKE

    Last Updated:Apr 15, 2026

    The DeepfakeDetectIntl and DeepfakeDetectIntlStream API operations analyze input face images or videos to detect deepfake risks, such as AI-generated faces, deepfake face swaps, template attacks, and presentation attacks. The response includes risk tags and confidence scores.

    API information

    The deepfake face detection service analyzes input face images or videos to identify potential deepfake threats. It provides two API operations to support different data submission methods:

    DeepfakeDetectIntlStream

    This operation supports the direct upload of image or video data as a file stream. Use this operation when processing local files.

    • Request method: HTTPS POST

    • QPS limit: Each API has a dedicated QPS limit. For details, see QPS limits of ID Verification server-side APIs.

    • Endpoints:

      Note

      • Benefits of internal network access: An internal network enables private communication between Alibaba Cloud services in the same region. If your application server is also in the same region, use the internal network endpoint to access the ID Verification service for a more secure and stable connection. 

      • Optimization for overseas access: Network conditions outside the Chinese mainland can be complex. To reduce latency and minimize request failures, optimize your integration by following the best practices in Server-side Network Latency Analysis and Optimization.

      Singapore

      • Public endpoint: cloudauth-intl.ap-southeast-1.aliyuncs.com

      • Internal endpoint: cloudauth-intl-vpc.ap-southeast-1.aliyuncs.com

    DeepfakeDetectIntl

    Use this operation to submit image or video data using a public URL or as a Base64-encoded string.

    • Request method: HTTPS POST

    • QPS limit: Each API has a dedicated QPS limit. For details, see QPS limits of ID Verification server-side APIs.

    • Endpoints:

      Note

      • Benefits of internal network access: An internal network enables private communication between Alibaba Cloud services in the same region. If your application server is also in the same region, use the internal network endpoint to access the ID Verification service for a more secure and stable connection. 

      • Optimization for overseas access: Network conditions outside the Chinese mainland can be complex. To reduce latency and minimize request failures, optimize your integration by following the best practices in Server-side Network Latency Analysis and Optimization.

      Singapore

      • Public endpoint: cloudauth-intl.ap-southeast-1.aliyuncs.com

      • Internal endpoint: cloudauth-intl-vpc.ap-southeast-1.aliyuncs.com

    Online debugging and integration

    Note

    Before you debug or integrate, read the Use OpenAPI to debug and integrate server-side APIs guide to understand how to call APIs on the OpenAPI platform and how to obtain the SDK.

    Image and video requirements

    Note

    The following requirements apply to both the DeepfakeDetectIntlStream and DeepfakeDetectIntl operations.

    The deepfake detection algorithm works best when it is combined with multi-dimensional data, such as business context, device information, and user behavior. The algorithm is ideal for scenarios where you already use a face detection SDK and want to enhance forged-face detection. To ensure stable model performance, submit face images or videos that are captured by a front-facing camera at a normal angle and meet the following requirements:

    Images

    • Format: JPG or JPEG.

    • Size: The recommended size is 50 KB to 100 KB. The maximum size is 10 MB. For images larger than 1 MB, you must upload them using a URL.

    • Resolution: The recommended resolution is 640 × 480 pixels (height × width). The maximum resolution is 1920 × 1080 pixels (height × width). The height must be greater than the width. If you scale the image, scale the shorter side to 720 pixels and use a compression ratio of 0.8.

    • Quality: The image must be clear and have normal exposure. Avoid faces that are too dark, too bright, or have halos.

    • Multiple faces: If an image contains multiple faces, the system detects the largest face by default.

    Videos

    • Format: MP4 or MOV.

    • Size: The video must be larger than 100 KB. The maximum size is 10 MB. For videos larger than 1 MB, you must upload them using a URL.

    • Length: 1 to 3 seconds.

    • Resolution: The recommended resolution is 640 × 480 pixels (height × width).

    • Frame rate: 8 frames per second (fps) or higher.

    Note

    For videos, the service extracts 8 frames by default for processing by the Qwen large model. To ensure accurate detection, do not submit invalid videos.

    Request parameters

    Name

    Type

    Required

    Description

    Example

    ProductCode

    string

    Yes

    The product to integrate.

    Set the value to FACE_DEEPFAKE.

    FACE_DEEPFAKE

    SceneCode

    string

    Yes

    A custom authentication scenario ID. Use this ID to query related records in the console.

    The ID must be a 10-character string containing letters, digits, and underscores (_).

    1234567890

    MerchantBizId

    string

    Yes

    A unique identifier for the merchant request. It must be a 32-character alphanumeric string.

    The first few characters are a custom merchant abbreviation. The middle part can include a timestamp. The last part can be a random or incremental sequence.

    e0c34a77f5ac40a5aa5e6ed20c******

    FaceInputType

    string

    No

    The input type of the face material:

    • IMAGE (default): A face image.

    • VIDEO: A face video.

      Note

      Video processing takes longer. Set the timeout to more than 3 seconds.

    IMAGE

    FaceUrl

    string

    No

    • DeepfakeDetectIntl: Submit data using either FaceUrl or FaceBase64. Choose one.

    • DeepfakeDetectIntlStream: Submit data using FaceUrl, FaceBase64, or FaceFile. Choose one.

    The URL of the face image or video.

    https://cn-shanghai-aliyun-cloudauth-xxxxxx.oss-cn-shanghai.aliyuncs.com/verify/xxxxx/xxxxx.jpeg

    FaceBase64

    string

    The Base64-encoded face image. For videos, upload using a URL.

    /9j/4AAQSkZJRgABAQAASxxxxxxx

    FaceFile

    string

    Face image or video file stream.

    Response parameters

    Note

    Because the detection model was upgraded to a large model for classification, the response has changed as follows:

    • The Result field no longer returns the value 2 (suspicious).

    • The RiskTag field no longer includes SuspectPSFace.

    • When the RiskScore field indicates a risk, it returns a fixed value of 0.96.

    Name

    Type

    Description

    Example

    RequestId

    String

    The request ID.

    8FC3D6AC-9FED-4311-8DA7-C4BF47D9****

    Code

    String

    Status code.

    Success

    Message

    String

    A detailed description of the status code.

    success

    ResultObject

    Result

    String

    The risk result:

    • 0: Low risk

    • 1: High risk

    • 2: Suspicious

    1

    RiskTag

    String

    The risk tag. Multiple tags are separated by commas (,). Tags include the following:

    • SuspectDeepForgery: Suspected deepfake

    • SuspectPSFace: Suspected synthetic attack

    • SuspectTemple: Suspected template attack

    • SuspectRemake: Suspected presentation attack

    SuspectDeepForgery

    RiskScore

    Map

    A map of risk scores. Returns the corresponding tag when a risk is detected.

    {
 "SuspectPSFace": "0.9796",
 "SuspectRemake": "0.0000",
 "SuspectDeepForgery": "0.9500",
 "SuspectTemple":"0.9500"
 }

    TransactionId

    String

    The unique ID of the verification request.

    08573be80f944d95ac812e019e36****

    Status codes

    HTTP Status Code

    Code

    Description

    200

    Success

    The request was successful.

    400

    MissingParameter

    A parameter cannot be empty.

    400

    InvalidParameter

    The parameter is invalid.

    401

    NoFaceDetected

    Feature extraction failed for the face in the custom source image. Upload a different image.

    401

    UnqualifiedPhoto

    The uploaded image is unreadable or its resolution does not meet the requirements. Replace the image. Make sure that the photo is clear, properly exposed, complete, and unobscured, and that the subject's head has no significant tilt.

    401

    ToolargeImage

    The image is too large. Compress the image or use a different upload method.

    401

    DataDuplication

    You passed images using both Base64 and URL methods simultaneously.

    401

    DownloadTimeout

    The image download from the URL timed out.

    403

    Forbidden.RAMUserAccessDenied

    Grant the AliyunAntCloudAuthFullAccess permission to the RAM user. For more information, see Grant a RAM user permissions to access the service.

    403

    Forbidden.AccountAccessDenied

    Make sure that you have activated ID Verification and that your account does not have an overdue payment.

    403

    Throttling.Api

    The API call is blocked by throttling.

    500

    InternalError

    An internal system error occurred. Contact technical support.