Face Inclination Estimation for Detection Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional facial organ detection systems using deep learning are highly accurate for upright faces but experience reduced accuracy when detecting inclined faces, as they are not effectively trained to handle variations in face orientation.
Innovation Solution
An information processing apparatus that outputs evaluation values for multiple reference angles to estimate the inclination angle of a face in an image, allowing for adjusted detection processing to improve accuracy in detecting inclined faces by correcting the detection angle based on the estimated inclination.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If facial organ detection is trained only with upright images, then detection accuracy for upright faces is improved, but detection accuracy for inclined faces deteriorates
Solution Approach 1:
The system performs preliminary estimation of the face inclination angle using a direction estimator before conducting the actual facial organ detection. This preliminary action allows the system to prepare appropriate adjustment parameters in advance, enabling accurate detection even when faces are inclined at various angles.
Solution Approach 2:
The system changes the detection parameters based on the estimated inclination angle. By adjusting detection parameters according to the face orientation, the system maintains high detection accuracy across different face inclinations without requiring separate training for each angle.
2Adaptability or versatility
If multiple face direction estimators are used to determine face direction, then adaptability to different face orientations is improved, but device complexity increases
Solution Approach 1:
The detection process is segmented into two distinct stages: first, estimating the face inclination angle using a direction estimator, and second, performing facial organ detection with adjusted parameters. This segmentation allows the system to achieve comprehensive angle coverage without requiring multiple simultaneous estimators for different directions.
Solution Approach 2:
The inclination angle estimation acts as an intermediary step between image input and facial organ detection. This intermediate processing provides crucial orientation information that enables the subsequent detection stage to adapt its parameters, effectively bridging the gap between fixed training conditions and variable real-world face orientations.
3Measurement precision
If detection processing is adjusted based on estimated inclination angle, then detection accuracy for inclined faces is improved, but processing time increases
Solution Approach 1:
The system performs only the necessary adjustment based on the estimated inclination angle rather than reprocessing the entire detection pipeline. By applying partial adjustment to the detection parameters, the system achieves improved accuracy for inclined faces while minimizing additional processing time.
Data Source
AI summary
There is provided with an information processing apparatus. An outputting unit, for each of a plurality of reference angles, outputs an evaluation value indicating whether a detection target in an image is inclined at the reference angle with respect to a standard orientation of the detection target. A first estimating unit estimates an inclination angle of the detection target in the image with respect to the standard orientation based on the evaluation values that have been respectively output for the plurality of reference angles. A detecting unit detects the detection target through processing in which an adjustment has been made using the estimated inclination angle.