Human Object Tracking With Head-Area and Body Similarity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In multi-object tracking scenarios, especially with human objects close to each other or with similar appearances, feature information confusion leads to object overlapping, reducing the accuracy and reliability of tracking results.
Innovation Solution
Utilize the head area of human bodies for tracking, combining feature information of the human body as a whole with detection frames of the head area, and employ a multi-task model to merge human body and head area detection tasks, reducing model complexity and data processing, and use similarity metrics to accurately match objects across frames.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If feature information of the entire human body is used for object matching, then the matching coverage is comprehensive, but the object matching accuracy deteriorates when multiple human objects are close or have similar appearances
Solution Approach 1:
The patent segments the human body into two parts: the entire human body and the head area. By using both the whole body detection frame and the head area detection frame separately, the system maintains comprehensive matching coverage while improving accuracy. The head area provides distinctive features for differentiation, while the whole body provides contextual information.
Solution Approach 2:
The patent introduces a new dimension by adding head area detection to the traditional whole-body detection. This additional dimensional information (head area detection frame) helps distinguish between similar objects without losing the comprehensive coverage provided by whole-body detection.
2Measurement precision
If separate models are used for human body detection and head area detection, then the detection accuracy is improved, but the device complexity increases
Solution Approach 1:
The patent merges the human body detection model and the head area detection model into a unified multi-task model. This single model performs both detection tasks simultaneously, reducing device complexity while maintaining the detection accuracy benefits of using both detection frames.
Solution Approach 2:
The multi-task model is designed to be universal, handling both whole-body detection and head area detection within a single framework. This multi-functional approach eliminates the need for separate models, thereby reducing system complexity while preserving detection accuracy.
3Measurement precision
If multiple detection frames are processed simultaneously, then the object matching accuracy is improved, but the data processing time increases
Solution Approach 1:
The patent performs preliminary action by using the multi-task model to generate both the whole body detection frame and head area detection frame in a single pass during the detection phase. This preliminary generation of both frames simultaneously reduces the need for separate processing steps later, thereby reducing overall data processing time while maintaining high matching accuracy.
Data Source
AI summary
An object tracking method, and a terminal device and a computer-readable storage medium using the same are provided. The method includes: obtaining first feature information of a target human body in a first image and a first detection frame of a head area of the target human body; obtaining second feature information of each human object in a second image and a second detection frame of a head area of the human object by performing a first image detection on the second image; and recognizing the target human body from the human object in the second image according to a first similarity between the first feature information and the second feature information and a second similarity between the first detection frame and the second detection frame. The above-mentioned method can effectively improve the accuracy of object matching, thereby enhancing the reliability of the results of multi-object tracking.