Dynamic Face Detection Threshold for Varying Shooting Scenes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image pickup systems use a fixed threshold for face detection, which is not suitable for varying shooting scenes, such as business portraits and group photographs, leading to inconsistent face recognition accuracy.
Innovation Solution
An image processing apparatus and method that allows users to set and adjust the face detection threshold based on the specific shooting scene, enabling tailored face region detection for different photography scenarios.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a fixed threshold of detection likelihood is used for face region detection, then the detection process is simple and fast, but the face recognition accuracy deteriorates when applied to different shooting scenes
Solution Approach 1:
The patent applies dynamics by making the detection threshold variable rather than fixed. The threshold is dynamically adjusted based on the shooting scene type (e.g., business portrait, group photograph, close-up), allowing the system to optimize face detection accuracy for different scenarios while maintaining reasonable processing speed through automated threshold selection.
Solution Approach 2:
The patent changes the parameter of detection threshold based on shooting scene characteristics. Different shooting scenes have different requirements for face detection sensitivity, and the system modifies the threshold parameter accordingly to balance between detection accuracy and processing efficiency for each specific scene type.
2Measurement precision
If a high threshold of detection likelihood is used, then face recognition accuracy for business portraits is improved, but the ability to detect all persons in group photographs deteriorates
Solution Approach 1:
The patent applies local quality by tailoring the detection threshold to specific local contexts (shooting scenes). Instead of using a single uniform threshold for all scenarios, the system selects different threshold values appropriate for each shooting scene type, such as higher thresholds for business portraits where accuracy is critical and lower thresholds for group photographs where completeness is prioritized.
Solution Approach 2:
The system dynamically adapts the detection threshold based on the identified shooting scene. The threshold is not static but changes according to the photographic context, enabling the same detection system to optimize for different objectives across various shooting scenarios without manual intervention.
3Adaptability or versatility
If a low threshold of detection likelihood is used, then the ability to detect all persons in group photographs is improved, but face recognition accuracy for business portraits deteriorates
Solution Approach 1:
The patent applies local quality by selecting different threshold levels appropriate for each shooting scene type. For group photographs, a lower threshold is used to ensure all persons are detected even if individual face quality varies. For business portraits, a higher threshold ensures only high-quality face detections are made, filtering out false positives.
Data Source
AI summary
In an image processing apparatus, a shooting scene and shooting parameters for each image pickup apparatus that is to be used to pick up an image are set in accordance with a user input. A picked-up image is received from an image pickup apparatus. In accordance with the shooting scene that is set for the image pickup apparatus which has picked up the received image, a threshold of the likelihood of detecting a face region that is included in the shooting parameters for the shooting scene is obtained. A face region is detected from the picked-up image using the obtained threshold of the likelihood of detecting a face region. An image of the detected face region is highlighted and displayed on a display screen.