Dashboard Camera Event Filtering Using Sensor-Triggered Video Upload
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional dashboard camera systems face high processing loads due to large amounts of image data, leading to increased data communication costs and unnecessary video data transmission.
Innovation Solution
The system acquires video and sensor data from a vehicle, identifies events based on sensor data, and transmits only relevant video data, reducing processing load and data communication by using machine learning models trained on smaller sensor data sets.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image data is used for event identification, then the accuracy of event detection is improved, but the processing load and data communication cost increase significantly
Solution Approach 1:
The system segments the data processing task by using sensor data for initial event identification and only using image data when necessary (when sensor data is insufficient or ambiguous). This divides the heavy image processing workload into selective cases, reducing overall processing load while maintaining detection accuracy for significant events.
Solution Approach 2:
Sensor data acts as an intermediary that filters and pre-identifies events before image data processing is triggered. The sensor data processing serves as a intermediate step that reduces the volume of data requiring complex image analysis, thereby lowering processing load while preserving detection capability.
2Loss of information
If all video data is transmitted for analysis, then complete information is available for event verification, but data communication cost and bandwidth consumption increase
Solution Approach 1:
The system extracts and transmits only the specific video segments that contain actual events of interest, rather than transmitting all video data. By using sensor data to identify event timestamps and extracting only those relevant video portions, the system minimizes data communication cost while ensuring complete information about actual events is preserved.
Solution Approach 2:
The transmission parameter changes from continuous video streaming to event-based selective transmission. The system changes the transmission trigger from time-based (continuous) to event-based (discrete), transmitting video data only when sensor data indicates an event has occurred, thereby reducing communication cost while maintaining information completeness for relevant events.
3Measurement precision
If image data is used for neural network processing, then event classification accuracy is improved, but the processing time and computational resources increase
Solution Approach 1:
The system performs preliminary event detection using sensor data before triggering image-based neural network processing. This preliminary action filters out non-event periods, so the computationally intensive image processing is only performed when actually needed, reducing overall processing time while maintaining classification accuracy for genuine events.
Solution Approach 2:
Instead of continuous image processing, the system uses periodic event-triggered processing based on sensor data thresholds. The neural network processes image data periodically only when sensor data indicates an event condition, converting continuous processing into discrete periodic operations that reduce total processing time while maintaining accuracy.
Data Source
AI summary
A dashboard camera acquires video data from a camera that captures the surroundings of a vehicle, acquires sensor data including at least one of acceleration, speed, and angular velocity of the vehicle, gives time information indicating time of acquisition to each of the video data and the sensor data, and transmits the sensor data to a data analysis server when an event occurs in the vehicle, the data analysis server identifies the content of the event occurred in the vehicle based on the sensor data and transmits an identification result to the dashboard camera, and when the identified event is determined to be the predetermined event, the dashboard camera specifies video data to which time information same as the time information given to the sensor data used for the identification is given, and transmits the specified video data to a video accumulation server.