Object Area Tracking Apparatus for Backward Subject Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object detection systems face challenges in accurately identifying and tracking objects, particularly when the face is not sufficiently visible, leading to reduced detection rates and incorrect main object selection, especially in scenarios where subjects are partially or fully backward.
Innovation Solution
An object area tracking apparatus that employs multiple detection methods, including face and body detection units, integrates detection results, and a main object determination unit to prioritize and select the main object based on past detection data, ensuring accurate tracking and selection even when the face is not directly detectable.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If face detection method is used to detect objects, then detection precision is improved when face is visible, but detection reliability deteriorates when person is backward or face is not sufficiently obtained
Solution Approach 1:
The system employs multiple detection units with different detection capabilities (face detection unit and body detection unit) to detect objects under various conditions. The face detection unit detects objects when the face is visible, while the body detection unit detects objects when the person is backward or face is not sufficiently obtained. This multi-functional approach ensures reliable detection regardless of the object's orientation or visibility.
Solution Approach 2:
The system changes the detection parameters by switching between different detection methods based on the detection situation. When the face is not sufficiently obtained, the system transitions from face-based detection parameters to body-based detection parameters, allowing continuous and reliable object detection under varying conditions.
2Reliability
If multiple detection methods are used to improve detection rate, then detection reliability is improved, but device complexity increases
Solution Approach 1:
The system merges the results from multiple detection units (face detection unit and body detection unit) through a detection result integration unit. This integration consolidates the detection results into a unified output, maintaining high detection reliability while managing system complexity through coordinated operation of specialized components.
Solution Approach 2:
The main object determination unit performs multiple functions by determining the main object based on detection results from different detection units. It evaluates detection situations, integrates results, and selects the primary object for tracking, thereby managing the complexity introduced by multiple detection methods through a centralized decision-making component.
3Reliability
If body detection is used to detect objects when face is not visible, then detection reliability is improved, but main object determination accuracy deteriorates due to incorrect selection
Solution Approach 1:
The main object determination unit uses feedback from the detection result integration unit to determine the main object. It considers the detection situation, the type of detection unit that detected the object, and integration results to accurately identify the main object even when body detection is used, thereby maintaining high determination accuracy despite relying on alternative detection methods.
4Stability of the object's composition
If detection results from multiple units are integrated, then detection stability is improved, but processing complexity increases
Solution Approach 1:
The detection result integration unit merges detection results from multiple detection units by consolidating their outputs into a unified detection situation. This merging process stabilizes the detection by providing a comprehensive view of all detection inputs while managing processing complexity through systematic integration of results.
Data Source
AI summary
An object area tracking apparatus has: a face detection unit for detecting a face area on the basis of a feature amount of a face from a supplied image; a person's body detection unit for detecting an area of a person's body on the basis of a feature amount of the person's body; and a main object determination unit for obtaining a priority for each of the objects by using detection results by the face detection unit and the person's body detection unit and determining a main object of a high priority, wherein for the object detected only by the person's body detection unit, the priority is changed in accordance with a past detection result of the object in the face detection unit.