Object Detection via Multi-View Signatures and Device Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video surveillance systems face challenges in identifying and tracking objects of interest across multiple fields of view, especially in densely populated areas and under varying conditions such as changes in lighting or camera settings, and often fail to accurately locate objects using signals from communication devices.
Innovation Solution
A system that generates signatures for objects of interest based on characteristics from multiple camera views, uses MAC addresses and RSSI values to associate communication devices with objects, and tracks these devices by mapping RSSI values to camera fields of view, enabling accurate location and identification across different locations and conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If video surveillance systems use search algorithms for single camera field of view, then identification accuracy is improved, but the system is unable to process multiple FOV inputs
Solution Approach 1:
The system creates a universal object representation that can be matched across multiple camera fields of view. By generating a signature from video content and matching it against object database entries, the system achieves consistent identification across different FOVs, making the search algorithm universally applicable to multi-camera environments.
Solution Approach 2:
The patent introduces an intermediary object database that stores signatures of objects of interest. This database acts as a mediator between multiple camera inputs and the search algorithm, enabling the system to process multiple FOVs by comparing video content against stored object signatures rather than directly coordinating multiple camera feeds.
2Adaptability or versatility
If known systems assume clear overlaps between FOV's for multi-FOV processing, then processing capability is improved, but the system fails in real-world scenarios without clear overlaps
Solution Approach 1:
The system extracts object signatures from video content independently of camera position or FOV overlap. By taking out the essential identifying features and storing them in an object database, the system can match objects across non-overlapping FOVs without requiring spatial coordination between cameras.
Solution Approach 2:
The patent transitions from spatial-based matching (relying on FOV overlaps) to feature-based matching (using object signatures). This dimensional change from spatial coordinates to feature space enables the system to identify objects across any FOV configuration, including non-overlapping scenarios.
3Speed
If tracking-based systems are used for object identification, then real-time tracking is improved, but the system is prone to fail in densely populated areas
Solution Approach 1:
The system creates a copy of the object identification process by storing signatures in a database. Instead of continuously tracking objects through crowded scenes, the system copies the identification logic to match video content against stored signatures, maintaining reliability in densely populated areas where tracking would fail.
4Adaptability or versatility
If video surveillance systems operate under varying conditions (lighting, angles, camera settings), then environmental adaptability is improved, but identification accuracy deteriorates
Solution Approach 1:
The system handles parameter changes by creating signatures that are invariant to lighting, angle, and camera setting variations. By transforming video content into standardized object signatures, the system maintains identification accuracy despite changes in environmental parameters such as lighting conditions or camera angles.
Data Source
AI summary
A system and method for detecting an object of interest. An embodiment of a system or method may include receiving an indication an object of interest is present in a first area; obtaining, at the first area, a first set of unique characteristics of a respective set of mobile communication devices; receiving an indication the object of interest is present in a second area; obtaining, at the second area, a second set of characteristics of a respective second set of mobile communication devices; and associating a unique characteristic of a mobile communication device with the object of interest based on the first and second sets of characteristics.