AI Driving Event Detection Using Pre-Trained Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current AI systems for determining driving events in vehicles require extensive driving data and are inefficient in obtaining this data with minimal expense, limiting their effectiveness in real-time applications.
Innovation Solution
A method and electronic device using multiple trained models to analyze video sequences from vehicles, recognizing object locations and determining driving events by analyzing sequential changes, enabling efficient data collection and real-time event detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If extensive driving data is collected to improve driving event recognition accuracy, then the recognition rate improves, but the cost and time for data collection increases significantly
Solution Approach 1:
The system performs preliminary actions by pre-training multiple specialized models (object detection model, trajectory prediction model, event determination model) offline before actual driving event recognition. This allows the system to have ready-to-use recognition capabilities without needing to collect and process extensive data in real-time, thus improving recognition accuracy while reducing data collection time and costs.
Solution Approach 2:
The system creates simplified copies or representations of complex driving scenarios through trained models. Instead of collecting and storing vast amounts of raw driving data, the system uses trained models that capture the essential patterns and relationships, enabling accurate event recognition with minimal data collection requirements.
2Measurement precision
If multiple trained models are used to analyze video sequences and determine driving events, then the recognition rate and accuracy improve, but the device complexity increases
Solution Approach 1:
The system segments the complex driving event recognition task into multiple specialized sub-tasks, each handled by a dedicated model: object detection model for identifying objects, trajectory prediction model for predicting movement paths, and event determination model for final event classification. This segmentation improves accuracy for each specific task while making the overall system more manageable and interpretable despite the multiple components.
Solution Approach 2:
The system uses a multi-functional architecture where trained models process various types of input data (video sequences, object locations, trajectory information) and can determine multiple types of driving events (pedestrian crossing, vehicle collision, object falling). This universal approach allows a single system to handle diverse recognition tasks, improving overall accuracy without proportionally increasing complexity.
3Speed
If real-time analysis of video sequences is performed to detect driving events, then the timeliness of event detection improves, but the computational resources and processing time required increases
Solution Approach 1:
The system performs preliminary processing by pre-training models offline and preparing prediction frameworks before real-time operation. During actual driving, the pre-trained models can quickly process video sequences and determine events without requiring extensive real-time computation, thus achieving fast event detection while reducing real-time energy consumption.
Data Source
AI summary
Provided are an artificial intelligence (AI) system configured to simulate functions of a human brain, such as recognition, determination, or the like, using a machine learning algorithm, such as deep learning, and an application of the AI system. An electronic device includes: a processor; and a memory storing instructions executable by the processor, wherein the processor is configured to execute the instructions to cause the electronic device to: obtain, from a vehicle, a video sequence including a plurality of frames captured while driving the vehicle, recognize a location of an object included in at least one of the plurality of frames, analyze a sequential change with respect to the location of the object in the plurality of frames, and determine whether a driving event of the vehicle occurs.