Situational Awareness Guardian System for Incident Response
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing number of video feeds from surveillance cameras in metropolitan areas makes it difficult for governmental agencies to review them in real-time, leading to a decrease in situational awareness and the ability to identify potential dangerous situations, as existing algorithms are compute-power intensive and slow.
Innovation Solution
A system that receives audio and video from a user, detects instructions, identifies available cameras with a view of the user, and applies object and action recognition processing to determine compliance with commands, taking non-compliance actions when necessary.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all video feeds from surveillance cameras are reviewed in real-time, then situational awareness and ability to identify dangerous situations improve, but the complexity and computational resources required become unmanageable
Solution Approach 1:
The system segments the video monitoring task by dividing it into two stages: first, a lightweight algorithm filters video feeds to identify only those containing potential dangerous situations; second, a more sophisticated algorithm analyzes only the sub-selected feeds. This segmentation resolves the contradiction by making the overall system manageable while maintaining high situational awareness.
Solution Approach 2:
The system extracts and processes only the essential information from video feeds using a first algorithm that identifies potential dangerous situations. By taking out only the relevant video feeds for further analysis, the system reduces complexity while preserving the ability to identify dangerous situations.
2Measurement precision
If sophisticated object and action recognition algorithms are applied to all video feeds, then identification accuracy of dangerous situations improves, but processing speed decreases making real-time notification impossible
Solution Approach 1:
The analysis process is segmented into two sequential stages: a first algorithm performs rapid filtering to identify potential dangerous situations, then a second, more sophisticated algorithm performs detailed analysis only on the sub-selected feeds. This segmentation enables both high accuracy and real-time processing speed.
Solution Approach 2:
The system applies full computational analysis only to a partial subset of video feeds that are most likely to contain dangerous situations. By applying excessive processing only where needed rather than uniformly across all feeds, the system achieves high identification accuracy while maintaining real-time processing capability.
3Productivity
If computational resources are increased to process all video feeds with advanced algorithms, then real-time dangerous situation detection becomes possible, but the cost and resource requirements become unsustainable
Solution Approach 1:
The system segments computational resource allocation by assigning lightweight algorithms to filter all video feeds and reserving intensive computational resources for analyzing only the sub-selected feeds. This segmentation enables real-time detection capability while keeping overall resource consumption sustainable.
Solution Approach 2:
Computational resources are applied partially and selectively only to video feeds that have been pre-identified as potentially dangerous. By avoiding excessive computation on all feeds and applying it only where necessary, the system achieves real-time detection with sustainable resource consumption.
Data Source
AI summary
A process for improving situational awareness at an incident scene includes first receiving audio or video of a user and detecting, in one or both of the audio and video, an instruction directed to another user. A compliance metric associated with the instruction is then accessed and one or more available second cameras having a field of view that incorporates a current location of the another user are identified. One or more video streams from the second cameras are received that include the another user. An action of the another user is identified from the video streams and is correlated with the compliance metric to determine a level of compliance of the another user with the instruction. In response to determining, as a function of the correlating, that the level of compliance falls below a threshold level of compliance, a computing device taking a responsive noncompliance action.