Sensor Profile Aggregation for People Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing people detection systems using overhead cameras face challenges such as occlusion issues, duplicate counting, and missed detections due to reliance on single-image analysis and motion features, which are prone to noise and light changes, and struggle to accurately track individuals, especially those in close proximity or with minimal motion.
Innovation Solution
A cloud-based aggregator system that utilizes multiple sensors to receive profiles of movement attributes, computes similarity measures between profiles from different sensors, and generates a global profile to accurately count and track objects, reducing errors by considering richer real-time motion information and allowing for privacy preservation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single overhead sensor is used for people detection, then the system complexity is reduced, but the detection accuracy and reliability deteriorate due to occlusion issues and limited viewing angles
Solution Approach 1:
The patent combines multiple sensors (first sensor and second sensor) to detect the same object from different positions. By merging the detection results from multiple sensors, the system overcomes the limitations of single-sensor detection including occlusion issues and limited viewing angles, thereby improving detection reliability without requiring complex installation of a single omnidirectional sensor
Solution Approach 2:
The patent transitions from single-dimension (single sensor position) to multi-dimension (multiple sensor positions) detection. By placing sensors at different locations and combining their detection results through similarity computation, the system achieves comprehensive coverage and improved reliability without the complexity of a single complex sensor installation
2Reliability
If multiple sensors are deployed to improve detection accuracy, then the reliability of object detection improves, but the device complexity and computational requirements increase
Solution Approach 1:
The patent extracts only the essential movement attributes from each sensor's detection data to create compact profiles. By taking out only the relevant features (movement attributes) rather than processing complete sensor data, the system reduces computational complexity while maintaining detection reliability through similarity comparison of these extracted profiles
Solution Approach 2:
The patent transforms raw sensor data into standardized profiles with specific movement attributes. By changing the parameter representation from raw sensor output to normalized movement attribute profiles, the system enables efficient similarity computation and reduces the complexity of comparing data from multiple sensors
3Productivity
If motion features are used for tracking, then the tracking capability is improved, but the measurement precision deteriorates due to noise and light changes
Solution Approach 1:
The patent combines movement attributes from multiple sensors to create a more robust tracking system. By merging data from multiple independent sensors, the system compensates for noise and light changes affecting individual sensors, improving measurement precision while maintaining tracking capability through the combined information
Solution Approach 2:
The patent uses similarity computation as a feedback mechanism to verify tracking accuracy. By continuously computing similarity between profiles from different sensors and comparing against threshold values, the system provides feedback on tracking precision and can adjust or reinitialize tracking when precision deteriorates due to noise or light changes
4Speed
If single-image analysis is used for detection, then the processing speed is improved, but the measurement precision deteriorates due to occlusion and limited information
Solution Approach 1:
The patent performs preliminary actions by capturing images at multiple time points before final analysis. By taking multiple images at different timestamps and combining their movement attributes, the system gathers more information about object characteristics and position, improving detection precision while maintaining processing speed through efficient profile aggregation
Data Source
AI summary
This disclosure describes techniques to aggregate sensor data. The techniques perform operations comprising: receiving, from a first sensor, a first profile representing a first set of movement attributes detected by the first sensor in an area at a given point in time; receiving, from a second sensor, a second profile representing a second set of movement attributes detected by the second sensor in the area at the given point in time; computing a similarity measure between the first and second sets of movement attributes of the first and second profiles; determining that the similarity measure exceeds a threshold value; and in response to determining that the similarity measure exceeds the threshold value, associating the first and second profiles with a same first object that is in the area at the given point in time.