Object Tracking via Segmented Surveillance Data Retrieval
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In distributed surveillance data storage environments, retrieving and analyzing large amounts of surveillance data from multiple devices is inefficient due to limited network bandwidth and high costs, making real-time analysis of surveillance data for object tracking challenging, especially in scenarios like traffic accident or criminal investigations.
Innovation Solution
An object tracking method that determines initial object spots based on location and time, retrieves and analyzes segments of surveillance data, and matches metadata against target object qualifications, reducing the amount of data needed for analysis by focusing on specific segments and locations, thereby minimizing data retrieval and analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all surveillance data from multiple devices is retrieved and analyzed, then object tracking completeness is improved, but network bandwidth consumption increases and processing efficiency decreases
Solution Approach 1:
The patent divides the surveillance data retrieval process into segments by identifying specific time ranges and spatial regions of interest. Instead of retrieving all data from all devices, the system segments the search space based on initial object detection locations and predicted movement patterns, retrieving only relevant data segments from specific surveillance devices during specific time periods.
Solution Approach 2:
The patent extracts and prioritizes key metadata from surveillance data for quick filtering. By extracting metadata such as object detection results, timestamps, and location information, the system can quickly identify which data segments warrant full analysis, avoiding the need to process entire datasets from all surveillance devices.
2Device complexity
If surveillance data is collected to a central location for analysis, then centralized processing capability is improved, but network bandwidth requirements increase
Solution Approach 1:
The patent implements local quality by enabling surveillance devices or local processing units to perform preliminary analysis and filtering of data before transmission to central systems. Each surveillance node processes its own data locally to identify objects of interest, extracting only relevant information for centralized analysis, thereby reducing network bandwidth requirements while maintaining centralized processing capabilities.
3Speed
If large amounts of surveillance data are retrieved in real-time, then tracking speed is improved, but network bandwidth bottlenecks worsen
Solution Approach 1:
The patent applies preliminary action by pre-processing surveillance data to extract and store metadata information such as object detection results, timestamps, and location data before full analysis is required. When tracking is needed, the system quickly queries this pre-processed metadata to identify relevant data segments, avoiding the need to scan and analyze entire surveillance datasets in real-time, thus achieving fast tracking with reduced data transmission.
Data Source
AI summary
A method for object tracking is provided, which is suitable for retrieving and analyzing distributed surveillance data. The method for object tracking includes the following steps: determining a set of surveillance data corresponding to at least one initial object spot in a set of initial object spots according to a location and a time of the initial object spot; retrieving segments of surveillance data in the set of surveillance data; finding at least one discovered object spot matching a target object qualification in the set of surveillance data and adding the discovered object spot into a set of discovered object spots; setting the set of initial object spots to be the set of discovered object spots and repeating the aforementioned steps when the set of discovered object spots is not empty; and outputting the discovered object spot when the set of discovered object spots is empty.