Specify a face detection model

Artikkeli
08/28/2024

This guide shows you how to specify a face detection model for the Azure AI Face service.

The Face service uses machine learning models to perform operations on human faces in images. We continue to improve the accuracy of our models based on customer feedback and advances in research, and we deliver these improvements as model updates. Developers can specify which version of the face detection model they'd like to use; they can choose the model that best fits their use case.

Read on to learn how to specify the face detection model in certain face operations. The Face service uses face detection whenever it converts an image of a face into some other form of data.

If you aren't sure whether you should use the latest model, skip to the Evaluate different models section to evaluate the new model and compare results using your current data set.

Prerequisites

You should be familiar with the concept of AI face detection. If you aren't, see the face detection conceptual guide or how-to guide:

Evaluate different models

The different face detection models are optimized for different tasks. See the following table for an overview of the differences.

Model	Description	Performance notes	Landmarks
detection_01	Default choice for all face detection operations.	Not optimized for small, side-view, or blurry faces.	Returns face landmarks if they're specified in the detect call.
detection_02	Released in May 2019 and available optionally in all face detection operations.	Improved accuracy on small, side-view, and blurry faces.	Doesn't return face landmarks.
detection_03	Released in February 2021 and available optionally in all face detection operations.	Further improved accuracy, including on smaller faces (64x64 pixels) and rotated face orientations.	Returns face landmarks if they're specified in the detect call.

Attributes are a set of features that can optionally be detected if they're specified in the detect call:

Model	accessories	blur	exposure	glasses	headPose	mask	noise	occlusion	qualityForRecognition
detection_01	✔️	✔️	✔️	✔️	✔️		✔️	✔️	✔️ (for recognition_03 or 04)
detection_02
detection_03		✔️	✔️	✔️	✔️	✔️		✔️	✔️ (for recognition_03 or 04)

The best way to compare the performances of the detection models is to use them on a sample dataset. We recommend calling the Detect API on a variety of images, especially images of many faces or of faces that are difficult to see, using each detection model. Pay attention to the number of faces that each model returns.

Detect faces with specified model

Face detection finds the bounding-box locations of human faces and identifies their visual landmarks. It extracts the face's features and stores them for later use in recognition operations.

When you use the Detect API, you can assign the model version with the detectionModel parameter. The available values are:

detection_01
detection_02
detection_03

A request URL for the Detect REST API looks like this:

https://westus.api.cognitive.microsoft.com/face/v1.0/detect?detectionModel={detectionModel}&recognitionModel={recognitionModel}&returnFaceId={returnFaceId}&returnFaceAttributes={returnFaceAttributes}&returnFaceLandmarks={returnFaceLandmarks}&returnRecognitionModel={returnRecognitionModel}&faceIdTimeToLive={faceIdTimeToLive}

If you are using the client library, you can assign the value for detectionModel by passing in an appropriate string. If you leave it unassigned, the API uses the default model version (detection_01). See the following code example for the .NET client library.

string imageUrl = "https://raw.githubusercontent.com/Azure-Samples/cognitive-services-sample-data-files/master/Face/images/detection1.jpg";
var response = await faceClient.DetectAsync(new Uri(imageUrl), FaceDetectionModel.Detection03, FaceRecognitionModel.Recognition04, returnFaceId: false, returnFaceLandmarks: false);
var faces = response.Value;

Add face to Person with specified model

The Face service can extract face data from an image and associate it with a Person object through the Add Person Group Person Face API. In this API call, you can specify the detection model in the same way as in Detect.

See the following .NET code example.

// Create a PersonGroup and add a person with face detected by "detection_03" model
string personGroupId = "mypersongroupid";
using (var content = new ByteArrayContent(Encoding.UTF8.GetBytes(JsonConvert.SerializeObject(new Dictionary<string, object> { ["name"] = "My Person Group Name", ["recognitionModel"] = "recognition_04" }))))
{
    content.Headers.ContentType = new MediaTypeHeaderValue("application/json");
    await httpClient.PutAsync($"{ENDPOINT}/face/v1.0/persongroups/{personGroupId}", content);
}

string? personId = null;
using (var content = new ByteArrayContent(Encoding.UTF8.GetBytes(JsonConvert.SerializeObject(new Dictionary<string, object> { ["name"] = "My Person Name" }))))
{
    content.Headers.ContentType = new MediaTypeHeaderValue("application/json");
    using (var response = await httpClient.PostAsync($"{ENDPOINT}/face/v1.0/persongroups/{personGroupId}/persons", content))
    {
        string contentString = await response.Content.ReadAsStringAsync();
        personId = (string?)(JsonConvert.DeserializeObject<Dictionary<string, object>>(contentString)?["personId"]);
    }
}

string imageUrl = "https://raw.githubusercontent.com/Azure-Samples/cognitive-services-sample-data-files/master/Face/images/detection1.jpg";
using (var content = new ByteArrayContent(Encoding.UTF8.GetBytes(JsonConvert.SerializeObject(new Dictionary<string, object> { ["url"] = imageUrl }))))
{
    content.Headers.ContentType = new MediaTypeHeaderValue("application/json");
    await httpClient.PostAsync($"{ENDPOINT}/face/v1.0/persongroups/{personGroupId}/persons/{personId}/persistedfaces?detectionModel=detection_03", content);
}

This code creates a PersonGroup with ID mypersongroupid and adds a Person to it. Then it adds a Face to this Person using the detection_03 model. If you don't specify the detectionModel parameter, the API uses the default model, detection_01.

Note

You don't need to use the same detection model for all faces in a Person object, and you don't need to use the same detection model when detecting new faces to compare with a Person object (in the Identify From Person Group API, for example).

Add face to FaceList with specified model

You can also specify a detection model when you add a face to an existing FaceList object. See the following .NET code example.

using (var content = new ByteArrayContent(Encoding.UTF8.GetBytes(JsonConvert.SerializeObject(new Dictionary<string, object> { ["name"] = "My face collection", ["recognitionModel"] = "recognition_04" }))))
{
    content.Headers.ContentType = new MediaTypeHeaderValue("application/json");
    await httpClient.PutAsync($"{ENDPOINT}/face/v1.0/facelists/{faceListId}", content);
}

string imageUrl = "https://raw.githubusercontent.com/Azure-Samples/cognitive-services-sample-data-files/master/Face/images/detection1.jpg";
using (var content = new ByteArrayContent(Encoding.UTF8.GetBytes(JsonConvert.SerializeObject(new Dictionary<string, object> { ["url"] = imageUrl }))))
{
    content.Headers.ContentType = new MediaTypeHeaderValue("application/json");
    await httpClient.PostAsync($"{ENDPOINT}/face/v1.0/facelists/{faceListId}/persistedfaces?detectionModel=detection_03", content);
}

This code creates a FaceList called My face collection and adds a Face to it with the detection_03 model. If you don't specify the detectionModel parameter, the API uses the default model, detection_01.

Note

You don't need to use the same detection model for all faces in a FaceList object, and you don't need to use the same detection model when detecting new faces to compare with a FaceList object.

Next steps

In this article, you learned how to specify the detection model to use with different Face APIs. Next, follow a quickstart to get started with face detection and analysis.

Jaa