Face Image Detection via Head Pose Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Face swap technologies often result in inaccurate identification of face images, leading to potential portraiture right violations and the spread of false news, as they struggle to accurately replace fringe face areas, making it difficult to detect fake faces in images or videos.
Innovation Solution
A method and apparatus that perform facial keypoint detection, segment center and entire face area images, and estimate head poses to determine if a face object is fake by comparing differences in head pose information between the center and entire face area images, leveraging the characteristic that face fringe areas are stiff after swapping.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If face swap technology is used to replace faces in images or videos, then the face replacement can be performed automatically, but the fringe face area replacement accuracy deteriorates
Solution Approach 1:
The patent divides the face image into a center face area and a fringe face area based on facial keypoints. By segmenting the face into these two regions, the system can separately analyze and compare head pose information from each region, enabling detection of inconsistencies that indicate fake faces while maintaining automatic replacement capability.
2Productivity
If conventional face detection methods are used, then the processing speed is fast, but the detection accuracy of fake faces deteriorates
Solution Approach 1:
The system segments the face image into center and fringe areas, then performs head pose estimation on both segments. By comparing the head pose information from these segments, the system achieves higher detection accuracy for fake faces while maintaining efficient processing through parallel computation of pose estimates.
Solution Approach 2:
The patent introduces head pose information as an intermediary feature to detect fake faces. Instead of directly analyzing the visual similarity of replaced faces, the system uses head pose estimation as a mediator to uncover inconsistencies between center and fringe areas, thereby improving detection accuracy without sacrificing speed.
Data Source
AI summary
Embodiments of the present disclosure provide a method and apparatus for detecting a face image, electronic device, and computer readable medium. The method may include: performing facial keypoint detection on an image frame containing a face object; segmenting a center face area image and an entire face area image of a target face object from the image frame based on a facial keypoint detection result; performing head pose estimation respectively based on the center face area image and the entire face area image; and generating a detection result on whether the target face object contained in the image frame is a fake face object based on a difference between a piece of first head pose information estimated based on the center face area image and a piece of second head pose information estimated based on the entire face area image.