Face Estimation Example

Introduction

Face estimation covers face detection and facial landmark annotation. X-AnyLabeling supports models that create a face rectangle and five facial keypoints with a shared group_id, making it easy to keep each face and its landmarks linked as one object.

Supported Models

Model	Provider	Output
SCRFD_10G_KPS	InsightFace	Face rectangle + five landmarks
YOLOv6Lite_l-Face	MeiTuan	Face rectangle + five landmarks
YOLOv6Lite_m-Face	MeiTuan	Face rectangle + five landmarks
YOLOv6Lite_s-Face	MeiTuan	Face rectangle + five landmarks

Usage

1. Import your image files.
2. Select and load one of the face estimation models from the auto-labeling model list.
3. Click Run (i) to infer the current image, or use Ctrl+B to process all images in the current task.
4. Review the generated face rectangles and landmark points. Each detected face uses the same group_id for its rectangle and five points.

The generated landmarks are:

• left_eye
• right_eye
• nose or nost_tip, depending on the selected model
• left_mouth_corner
• right_mouth_corner

Export

Face estimation annotations use standard X-AnyLabeling shapes: a rectangle for each face and point shapes for facial landmarks. Export them with the format that matches your downstream training or conversion workflow.