Edge Image Surveillance Model Distribution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud-based analysis of edge device data for event detection sacrifices response time and increases bandwidth consumption, making edge-based analysis advantageous but not effectively implemented in existing systems.
Innovation Solution
A method and system that involve a management device identifying and transmitting event detection models to specific image capture devices within a defined geographic area and time frame, allowing these devices to process image data locally for event identification and notification, with the option to cease model usage after processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If cloud-based analysis is used for event detection, then processing requirements of edge devices are reduced, but response time is sacrificed and bandwidth consumption increases
Solution Approach 1:
The system pre-loads event detection models into edge device memory before they are needed for analysis. The management device identifies which image capture devices will be in the target geographic area during the analysis period and proactively transmits the necessary model data to those devices. This preliminary action ensures that when analysis begins, the edge devices already have the models in memory, enabling immediate local processing without cloud dependency and achieving fast response times while keeping edge device complexity manageable through selective model deployment.
Solution Approach 2:
The system implements differentiated processing by location and time - edge devices perform local event detection only when they are within the specified geographic area and time period, rather than continuously or universally. The management device transmits models selectively to devices that will be in the target area, enabling local processing capabilities to be activated precisely where and when needed, optimizing both response time and resource utilization.
2Device complexity
If cloud-based analysis is used for event detection, then processing requirements of edge devices are reduced, but bandwidth consumption increases
Solution Approach 1:
The system extracts only the necessary model data from the cloud and transmits it selectively to specific edge devices that will be in the target geographic area during the analysis period. Rather than transmitting all image data to the cloud or loading all possible models to every edge device, the management device identifies the subset of devices needed and sends only the relevant models to them. This extraction approach minimizes bandwidth consumption while enabling sufficient local processing capability.
Solution Approach 2:
The system implements partial action by transmitting models to only those edge devices that will be in the target geographic area during the analysis period, rather than to all edge devices. This selective transmission reduces unnecessary bandwidth consumption while providing sufficient processing capability where needed. The management device determines the subset of devices based on their projected locations and transmits models only to that subset.
3Loss of time
If edge-based analysis is implemented, then response time is improved, but processing requirements of edge devices increase
Solution Approach 1:
The system pre-loads event detection models into edge device memory before analysis is needed. The management device proactively transmits the necessary model data to edge devices that will be in the target geographic area, so that when analysis begins, the devices already have the models cached in memory. This eliminates the need for devices to download models during active analysis, reducing latency and enabling fast local processing without requiring devices to maintain permanent storage of multiple models.
Solution Approach 2:
The system implements dynamic model deployment where edge devices receive and cache models temporarily based on their projected location and the analysis time period. Devices dynamically load models into memory when needed for specific analysis tasks and can discard them afterward, allowing the system to provide high processing capability when needed while keeping average device complexity low. The management device coordinates this dynamic allocation based on real-time or predicted device locations.
Data Source
Figure 1A
Figure 1B
Figure 2A~2B
AI summary
Detection models are distributed as needed to a select subset of image capture devices, based on geographic location and time of the image capture devices. Each image capture device which receives a model processes image data according to the model, and provides a notification of any detected instances of an event the model is trained to detect. Distribution and detection can be based on historical data or live data.