Face Detection Judgment Area Adjustment for Profile and Frontal Views
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional face detection technologies suffer from high false positive rates, especially for profile face images, leading to incorrect detection of landscape images and reduced accuracy in identifying face images with varying directions and sizes.
Innovation Solution
A face detection device with a judgment area determining unit that adjusts the size and aspect ratio of judgment areas based on face direction, allowing for improved detectability by distinguishing between frontal and profile face images, and incorporating rotation ranges to account for different face orientations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If the judgment area size is increased to improve detection coverage, then the detection area is expanded, but the false positive rate increases
Solution Approach 1:
The patent applies local quality by differentiating judgment area sizes based on face direction. Frontal face images use smaller judgment areas while profile face images use larger judgment areas. This localized adaptation reduces false positives in frontal views and improves detection coverage in profile views, resolving the contradiction between detection area expansion and false positive reduction.
2Reliability
If the judgment area size is decreased to reduce false positives, then the false positive rate is reduced, but the detection coverage is limited
Solution Approach 1:
The patent implements local quality by selecting judgment area sizes according to face direction. Smaller judgment areas are used for frontal faces to reduce false positives, while larger judgment areas are used for profile faces to maintain detection coverage. This differentiated approach resolves the contradiction between false positive reduction and detection coverage maintenance.
3Device complexity
If a uniform judgment area size is used for all face directions, then the device complexity is reduced, but the detection accuracy for different face directions deteriorates
Solution Approach 1:
The patent applies local quality by configuring different judgment area sizes based on face direction (frontal vs. profile). This localized differentiation improves detection accuracy for each face direction while maintaining relatively simple device complexity through rule-based size selection rather than complex adaptive systems.
4Measurement precision
If the judgment area size is increased to detect profile face images, then the detection coverage for profile faces is improved, but the false positive rate for frontal faces increases
Solution Approach 1:
The patent resolves this contradiction by applying local quality through direction-specific judgment area sizes. Profile face detection benefits from larger judgment areas, while frontal face detection uses smaller judgment areas to minimize false positives. The system determines face direction and selects appropriate judgment area size accordingly, achieving optimal performance for each face type without compromising the other.
Data Source
AI summary
A face detection device comprises: a judgment area determining operable to judgment areas in an inputted image; and a face judgment unit operable to judge existence of a face image greater than a first predetermined size in the judgment areas. According to face directions, the judgment area determining unit determines a ratio of the inputted image to the judgment areas. The characteristics of face images can be effectively utilized.