FaceDetection

When to use the FaceDetection Activity

Use Face Detection to detect faces within images, and get back high-precision face bounding box and token for each detected face. It also allows you to store metadata of each detected face for future use.

Figure 1

Drag a Face Detection activity from the FacePlusPlus package and drop it into the FPP Scope. Specify the Image path based on your selection of Parameter 1 value. Fill appropriate details in all fields in the property window. Refer Figure 5. Output field supports only string variable type, while Result field supports only Boolean variable type. Create variables accordingly. Execute to detect faces within images, and get back high-precision face bounding box and token for each detected face.

Figure 2

Face Detection can detect all the faces within the image. Each detected face gets its face_token, which can be used in follow-up analysis and operations.

Return Values

Figure 3

Parameter 1

There are three options available in the dropdown and they are as follows. Explanation is provided in the ensuing image.

Figure 4

Figure 4.1

Technical Reference

Figure 5

Image Requirements

Format: JPG (JPEG), PNG
Size: between 48*48 and 4096*4096 (pixels)
File size: no larger than 2MB
Minimal size of face: the bounding box of a detected face is a square. The minimal side length of a square should be no less than 1/48 of the short side of image, and no less than 48 pixels. For example if the size of image is 4096 * 3200px, the minimal size of face should be 66 * 66px.

About face_token

If you want to use the detected face for follow-up operations, we recommend you add its corresponding face_token into FaceSet. A face_token expires within 72 hours if it is not added into any FaceSet. 

Every time you operate face detection upon one image, the face within the image will get a different face_token.