Surveillance Data Transfer for Model Updates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional surveillance systems face challenges in dynamically updating detection models for surveillance cameras due to limited network bandwidth, leading to increased erroneous detection and incomplete detection, especially when resources are scarce and network bandwidth is narrow.
Innovation Solution
An information transfer apparatus that selectively transmits data worth learning by assigning an index indicating the degree of worthiness based on a comparison with a reference model, allowing only necessary data to be transmitted for model updates, even in narrowband networks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all surveillance camera data is transmitted for model updates, then detection precision can be maintained, but network bandwidth requirements become prohibitively high
Solution Approach 1:
The patent extracts only the essential learning components (feature amounts and evaluation results) from the complete surveillance data, transmitting only these critical elements to the learning server. This selective extraction maintains detection precision while dramatically reducing the data volume that requires network transmission.
Solution Approach 2:
The surveillance data is segmented into distinct components: feature amounts extracted by the learning server, evaluation results generated by the camera's analysis unit, and metadata. This segmentation allows only the necessary segments to be transmitted over the network, rather than sending complete raw data.
2Reliability
If detection models are updated frequently to adapt to environmental changes, then surveillance reliability improves, but the workload and time required for updates increase
Solution Approach 1:
The learning server performs preliminary processing by extracting feature amounts from transmitted images before model updates are needed. This preliminary action prepares the data in advance, so that when environmental changes occur and model updates are required, the process can proceed quickly using pre-processed features rather than raw data.
Solution Approach 2:
The system implements feedback through the evaluation unit that continuously assesses detection accuracy and generates evaluation results. When detection precision drops below thresholds, this feedback triggers targeted model updates only when necessary, rather than continuous updates, reducing overall update time while maintaining reliability.
3Productivity
If surveillance cameras perform automatic surveillance with detection models, then manual surveillance workload decreases, but erroneous detection and incomplete detection increase over time
Solution Approach 1:
The evaluation unit continuously provides feedback on detection performance by comparing actual detections against ground truth data. When detection accuracy degrades due to environmental changes or target variations, this feedback triggers model updates, allowing the automatic surveillance system to self-correct and maintain high reliability without manual intervention.
Solution Approach 2:
The surveillance camera system performs self-updates by automatically transmitting data to the learning server, which generates updated detection models that are then downloaded back to the cameras. This self-service capability allows the system to adapt to changing conditions autonomously, maintaining detection accuracy while preserving automatic surveillance productivity.
Data Source
AI summary
A learning system (100) includes an information transfer apparatus (10) and a learning processing apparatus (20). The information transfer apparatus (10) includes: an analysis unit (11) that obtains data serving as a learning target, compares the obtained data with a reference model, and assigns, to the data, an index indicating a degree of worthiness of the data as the learning target; and a transmission processing unit (12) that transmits the data to the learning processing apparatus (20) based on a rule that has been set using the index. The learning processing apparatus (20) includes a learning processing unit (21) that updates the model or generates a new model based on the data transmitted from the information transfer apparatus (10).