Trajectory Map Visualization for Video Surveillance Pattern Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video surveillance systems require predefined definitions for objects and behaviors, making them labor-intensive and costly to maintain or adapt for different applications, and unable to recognize new patterns or changes in existing patterns.
Innovation Solution
A computer-implemented method that generates a display of information learned by a video surveillance system by receiving a request to view a trajectory map, retrieving trajectories of foreground objects, and superimposing their paths over a background image, using a combination of a computer vision engine and a machine learning engine to identify and recognize patterns by observing video frames.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If predefined definitions for objects and behaviors are used, then the system can recognize defined patterns, but the system becomes labor-intensive and costly to maintain or adapt
Solution Approach 1:
The surveillance system automatically learns and updates trajectory patterns by observing video frames without requiring manual programming. The system performs self-training by detecting objects, tracking their movements, and automatically incorporating new trajectory patterns into its knowledge base, eliminating the need for labor-intensive manual definition and maintenance of patterns
Solution Approach 2:
The system transitions from static predefined patterns to dynamic learned patterns that continuously adapt. Trajectory patterns are not fixed in advance but are dynamically discovered and updated based on observed object movements, allowing the system to automatically adjust to new behaviors and scenarios
2Reliability
If predefined definitions for objects and behaviors are used, then the system can recognize defined patterns, but the system cannot recognize new patterns or changes in existing patterns
Solution Approach 1:
The system automatically discovers and learns new trajectory patterns by monitoring object movements in video frames. When new movement patterns are detected, the system self-trains to recognize these patterns, enabling continuous adaptation to new behaviors without external intervention
Solution Approach 2:
The pattern recognition system evolves from static predefined patterns to dynamic learned patterns. The system continuously updates its understanding of normal and abnormal behaviors based on observed trajectories, enabling it to adapt to changing scenarios and recognize previously unknown patterns
3Adaptability or versatility
If separate software products are developed to recognize additional objects or behaviors, then recognition capabilities are expanded, but maintenance costs increase prohibitively
Solution Approach 1:
The surveillance system implements a universal learning mechanism that can recognize any trajectory pattern through automatic observation and learning, eliminating the need for separate specialized software products. The single system performs multiple recognition functions by learning diverse object behaviors and movement patterns uniformly
Solution Approach 2:
The system expands its recognition capabilities through self-learning rather than requiring additional software products. By automatically detecting and learning new trajectory patterns from video data, the system provides cost-effective expansion of recognition capabilities without external development costs
Data Source
AI summary
Techniques are disclosed for visually conveying a trajectory map. The trajectory map provides users with a visualization of data observed by a machine-learning engine of a behavior recognition system. Further, the visualization may provide an interface used to guide system behavior. For example, the interface may be used to specify that the behavior recognition system should alert (or not alert) when a particular trajectory is observed to occur.