In-Vehicle Video Summarization Using Multi-Condition Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems for creating summary moving images from in-vehicle camera data often result in monotonous content due to reliance on single extraction conditions, failing to capture a variety of driving situations.
Innovation Solution
An information processing apparatus that combines moving image data using multiple extraction conditions defined in a template, selecting and stitching scenes that satisfy various conditions such as object detection, sensor data, and vehicle location, to create a richly undulating summary moving image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple extraction conditions are used to select diverse scenes, then the richness and variety of the summary moving image is improved, but the device complexity and processing difficulty increase
Solution Approach 1:
The extraction conditions are segmented into multiple independent types (first extraction condition based on image recognition, second extraction condition based on sensor data, third extraction condition based on location information). Each condition type operates independently to detect different aspects of driving situations, allowing the system to achieve comprehensive scene selection without requiring complex integrated processing of all conditions simultaneously.
Solution Approach 2:
The system extracts moving image data based on partial satisfaction of extraction conditions rather than requiring all conditions to be met. When any of the multiple extraction conditions are satisfied, the system extracts and combines the corresponding moving image data, which simplifies the processing logic while still achieving diverse scene selection.
2Measurement precision
If multiple types of data are integrated for extraction conditions, then the accuracy of scene selection is improved, but the loss of time for processing increases
Solution Approach 1:
The system performs preliminary classification of extraction conditions into distinct types (image-based, sensor-based, location-based) that can be processed independently. By organizing the multiple data types into separate processing streams beforehand, the system can evaluate each condition type efficiently without requiring complex real-time integration of all data types, thus reducing processing time while maintaining selection accuracy.
Solution Approach 2:
Each extraction condition type operates autonomously to evaluate its specific data source and independently determine whether extraction should occur. The image recognition system, sensor data processor, and location information evaluator each self-service their respective conditions without requiring continuous coordination with the others, which significantly reduces processing time while maintaining accurate scene selection.
Data Source
AI summary
An information processing apparatus comprises controller comprising at least one processor configured to perform: obtaining a template for creating combined moving image data in which a plurality of moving image data are combined and, which includes information on at least one extraction condition defined based on a plurality of types of data, extracting a plurality of second moving image data that is moving image data in a predetermined period including a timing that conditions one or more the extraction conditions indicate are satisfied from first moving image data captured by an in-vehicle camera, creating the combined moving image data by combining the second moving image data extracted according to the template.