On-Board Image Abnormality Detection With Cloud Notification Split
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies for detecting abnormalities such as dirt in vehicle cabins lack a clear distribution of processes between the vehicle side and the cloud side, leading to inefficiencies, especially in vehicle sharing services where high convenience is required.
Innovation Solution
An abnormality detection system that includes a cloud for data collection and an on-board device connected to the cloud, utilizing multiple applications to analyze image data from a camera to detect abnormalities, with the on-board device transmitting analysis results and image data to the cloud for storage and notification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If image analysis is performed only in the cloud, then processing power requirements are reduced on the vehicle side, but communication data volume and notification time increase
Solution Approach 1:
The patent segments the abnormality detection system into on-board and cloud components. The on-board device performs initial image analysis and detects abnormalities, while the cloud performs secondary analysis and stores data. This segmentation enables fast local notification while maintaining overall system sophistication.
Solution Approach 2:
The on-board device performs preliminary abnormality detection and notification before cloud processing completes. This preliminary action ensures timely alerts to users and operators without waiting for full cloud analysis, resolving the time delay issue.
2Measurement precision
If multiple applications are executed on the on-board device, then detection accuracy improves, but processing load and energy consumption increase
Solution Approach 1:
Multiple detection applications are segmented between on-board and cloud devices. The on-board device runs essential detection apps for immediate abnormalities, while less critical analysis is performed in the cloud, reducing on-board energy consumption while maintaining comprehensive detection accuracy.
Solution Approach 2:
The system applies partial action by executing only the most critical detection applications on the resource-constrained on-board device, while allowing the cloud to handle additional analysis tasks that would otherwise increase on-board energy consumption.
3Loss of information
If all image data is transmitted to the cloud, then data availability for analysis improves, but communication bandwidth requirements and storage needs increase
Solution Approach 1:
The system extracts only essential information (abnormality detection results and key image data) for cloud transmission, rather than transmitting all captured images. This extraction approach maintains data availability for analysis while significantly reducing communication bandwidth requirements and cloud storage needs.
Solution Approach 2:
Instead of transmitting original high-resolution images, the system transmits processed copies containing only relevant abnormality information. This copying approach preserves essential data while minimizing data transmission volume and storage requirements.
Data Source
AI summary
An abnormality detection system includes a cloud and an on-board device communicatively connected to the cloud, and further includes multiple applications each configured to detect the abnormality as a target based on image data of an image captured by a camera. The detection unit is configured to: analyze the image data and detect the abnormality when each of the multiple applications is executed, transmit, to the cloud, an analysis result analyzed by the detection unit, transmit, to the cloud, at least the image data from which the abnormality is detected. The cloud includes a storage unit that stores the analysis result and the image data transmitted from the on-board device. The abnormality detection system notifies, using at least one of the on-board device or the cloud, a notification target of the analysis result.