Driving Record Video Annotation for Abnormal Event Retrieval
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for searching video clips from driving recorders are inefficient, as users typically estimate times to locate relevant footage, which is not accurate.
Innovation Solution
A data processing method that utilizes first driving data and historical driving features to annotate driving record videos, allowing for precise identification of abnormal vehicle behavior, thereby facilitating efficient video clip retrieval.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If users estimate approximate time to search for video clips, then the search process is simple, but the search efficiency is low and accuracy is poor
Solution Approach 1:
The system performs preliminary annotation of video clips with driving event information before the user needs to search. Driving events are detected and time stamps are recorded in advance, so when users need to search for specific video clips, they can directly access pre-annotated results rather than manually estimating time ranges.
Solution Approach 2:
The patent introduces an intermediary layer (annotation information) between the user and the video clip search process. This annotation layer contains driving event data and time stamps that mediate between the raw video footage and user search queries, enabling efficient retrieval without direct time estimation by users.
2Device complexity
If conventional time-based searching is used, then no additional processing is needed, but the accuracy of locating relevant footage is poor
Solution Approach 1:
The system performs preliminary detection and annotation of driving events with precise time stamps before search operations. This advance processing captures accurate temporal information about abnormal driving events, enabling precise video clip location when users search later.
Solution Approach 2:
The patent replaces manual time estimation (mechanical/user-based process) with automated driving event detection and timestamping (electronic/computational process). This substitution uses sensor data and algorithms to objectively determine event timing rather than relying on user approximation.
3Productivity
If driving events are annotated in video clips, then video clip search efficiency is improved, but data processing complexity increases
Solution Approach 1:
The patent segments the data processing into distinct modules: sensor data acquisition, driving event detection, timestamp generation, and video annotation. This segmentation allows each component to handle specific tasks independently, making the overall complex process more manageable and maintainable.
Solution Approach 2:
The system integrates multiple functions into a unified annotation process that simultaneously detects driving events, generates time stamps, and annotates video clips. This multi-functionality reduces the need for separate processing systems while achieving comprehensive video clip annotation.
Data Source
AI summary
Embodiments of this application provide a data processing method and apparatus, and a device. In an example method, first data and a first historical driving feature are obtained, where the first data includes first driving data of a vehicle in a first time period, and the first historical driving feature is determined based on historical driving data of the vehicle. It is determined, based on the first data and the first historical driving feature, that the vehicle is abnormal in the first time period. A driving record video of the vehicle based on the first time period is annotated.