Camera-Based Sports Timing with Depth-Zone Athlete Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vision-based sports timing systems face challenges in accurately and reliably determining the passing times and identities of participants in mass sports events, particularly due to the inclusion of non-participating individuals in the captured scene and privacy concerns related to biometric data usage.
Innovation Solution
A camera system with depth information and a virtual timing zone is used to filter out non-participating individuals by detecting objects within a calibrated 2D or 3D zone aligned with the track, applying object detection and feature analysis algorithms to identify and time only participating athletes, utilizing machine learning techniques for real-time processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If a camera system captures a wide scene around the timing line to ensure all participants are detected, then the detection coverage is improved, but the number of non-participating individuals (spectators, officials) included in the scene increases, making accurate identification more difficult
Solution Approach 1:
The patent divides the captured scene into multiple zones: a first zone containing only participating athletes (detected using pose estimation and sports uniform detection), and a second zone containing non-participating individuals. By segmenting the scene and applying different processing rules to each zone, the system maintains high identification accuracy while preserving comprehensive detection coverage.
Solution Approach 2:
The patent applies different quality standards and processing methods to different regions of the scene. Participants in the first zone undergo rigorous verification through pose estimation and uniform detection, while individuals in the second zone are excluded from timing results. This local quality approach ensures high reliability for participant identification without sacrificing overall scene coverage.
2Measurement precision
If traditional RFID timing systems are used to accurately time participants, then timing precision is improved, but the system requires providing UHF tags to every participant and is sensitive to environmental influences
Solution Approach 1:
The patent replaces the RFID mechanical/electromagnetic system with a vision-based system using cameras, pose estimation algorithms, and machine learning models. This substitution eliminates the need for UHF tags, removes sensitivity to environmental interference (moisture, rain, reflections), and reduces device complexity by using standard camera equipment combined with software processing.
Solution Approach 2:
The patent changes the fundamental measurement parameters from RFID signal detection to visual feature analysis. Instead of measuring electromagnetic signal presence, the system analyzes pose estimates, uniform colors, and spatial positions from camera images. This parameter change enables timing without physical tags and reduces environmental sensitivity.
3Productivity
If vision-based timing systems process all detected objects in the scene, then comprehensive detection is achieved, but computational resources are wasted on non-participating individuals and detection accuracy for participants decreases
Solution Approach 1:
The patent performs preliminary classification of detected objects into participants and non-participants using pose estimation and uniform detection before applying detailed timing analysis. This preliminary action filters out non-participants early in the processing pipeline, preventing waste of computational resources and maintaining high detection accuracy for actual participants.
Solution Approach 2:
The patent extracts and processes only the relevant subset of objects (participants in the first zone) from the complete scene. By taking out and separately processing participants using specialized algorithms (pose estimation, uniform detection), the system achieves high detection accuracy while improving processing efficiency by excluding non-participants from detailed analysis.
4Measurement precision
If biometric data is collected for identification to ensure accurate participant recognition, then identification accuracy is improved, but privacy concerns arise and data usage is restricted by regulations
Solution Approach 1:
The patent extracts and uses only non-biometric visual features for identification, such as sports uniform colors, patterns, and logos. By taking out the identification problem from the biometric domain and solving it through sport-specific visual features, the system achieves accurate participant recognition without collecting sensitive personal data, thereby eliminating privacy concerns and regulatory restrictions.
Data Source
AI summary
Methods and systems for determining a passing time of an object passing a timing line comprise receiving video frames captured by a calibrated camera system, each frame representing object(s) moving along a track; determining or receiving calibration data defining a virtual timing zone positioned at a distance from the camera system, the timing zone extending across and along the track and including a virtual timing line; detecting objects in the frames using an object detection algorithm to define detected objects; determining depth information for some of the video frames, the depth information comprising information regarding a distance between detected objects and the camera system; determining detected objects that are positioned within the timing zone based on the calibration data and the depth information; and, determining a passing time for detected objects in the virtual timing zone based on instance(s) of frame(s) comprising the detected object(s) passing the timing line.