Video Surveillance Tracking Human Object Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video monitoring systems in retail spaces and public gathering places face challenges in accurately distinguishing between humans and inanimate objects, failing to differentiate between various human subjects, and unable to track interactions or work schedules of employees, leading to inefficiencies in traffic flow analysis and conversion rate calculations.
Innovation Solution
A system comprising electronic video cameras and a central computer program that distinguishes human subjects from inanimate objects, identifies interactions, and correlates movements to associate individuals or groups with agents, using video and wireless identification data to track positions and behaviors, enabling accurate conversion rate calculations and employee monitoring.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If optical beam break people counters are used to monitor traffic flow, then counting capability is provided, but false counts occur due to interference from background light and inability to distinguish people from inanimate objects
Solution Approach 1:
The patent replaces optical beam break technology with video-based computer vision systems. The system uses cameras to capture visual data and algorithms to distinguish human subjects from inanimate objects, eliminating the false count problem inherent in optical beam methods while maintaining counting capability.
Solution Approach 2:
The system changes the detection parameter from optical beam interruption to visual pattern recognition. By analyzing visual parameters such as shape, movement patterns, and contextual information, the system accurately identifies human subjects without being interfered by background light or similar objects.
2Reliability
If video based systems are used for monitoring spaces, then false count problems are partially overcome, but the systems fail to distinguish between different classes of human subjects and cannot identify relationships between people
Solution Approach 1:
The patent segments the analysis into multiple levels: first distinguishing human from non-human, then categorizing human subjects into different classes (e.g., customers, employees), and finally identifying relationships between individuals. This hierarchical segmentation enables comprehensive information extraction from video data.
Solution Approach 2:
The system adds analytical dimensions to basic video monitoring. Beyond simple presence detection, it incorporates spatial relationships, temporal patterns, and contextual analysis to distinguish between different types of human subjects and identify their interactions, transforming raw video data into meaningful insights.
3Adaptability or versatility
If existing video systems are used, then basic monitoring is provided, but they are unsuitable for monitoring employee work schedules and cannot determine time spent on-site versus break time
Solution Approach 1:
The patent makes the video monitoring system multi-functional by enabling it to perform various tasks: counting customers, monitoring employee schedules, tracking work vs. break time, and analyzing traffic flow patterns. The same video infrastructure supports multiple business objectives through flexible software analysis modules.
Solution Approach 2:
The system continuously monitors employee positions and activities, providing real-time feedback on work schedules and time allocation. By analyzing temporal patterns and spatial data, the system can distinguish between productive work time and break time, enabling precise employee productivity management.
Data Source
AI summary
Embodiments of the invention may relate to systems and/or methods for using video surveillance systems to monitor spaces where people tend to gather such as retail stores, theaters, stadiums, or other public gathering places. Furthermore, embodiments may be adapted to discern elements of human behavior in a video feed, and use these behaviors to draw quantitative and/or qualitative conclusions from the video data. Typical conclusions may include overall conversion rates for a store, conversion/close rates of individual salespeople, traffic patterns within a space, peak traffic times, and so on.