Driver Behavior Detection Using Active Regions and Deep Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing deep learning-based object detection methods for detecting driver behaviors consume excessive computational resources, making them unsuitable for embedded vehicle environments, and fail to efficiently detect risky behaviors like smartphone use, smoking, and food taking during driving.
Innovation Solution
A driver behavior detection system using deep learning-based object classification that utilizes frame inputting, downsampling, active image production, active region extraction, and behavior detection units to sense brightness changes between frames, applying a deep learning-based object classification algorithm to detect driver behaviors with reduced computational requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a deep learning object detection algorithm is applied to detect driver behavior, then detection accuracy is improved, but computational resource consumption increases significantly
Solution Approach 1:
The patent divides the image processing task into two stages: first, a lightweight active region extraction algorithm identifies potential behavior regions using simple brightness comparison; second, the deep learning object detection algorithm is applied only to these extracted regions rather than the entire image. This segmentation approach maintains high detection accuracy while significantly reducing computational resource consumption by limiting the resource-intensive deep learning processing to small, relevant portions of the image.
Solution Approach 2:
The patent implements partial action by applying the deep learning algorithm only to active regions where driver behavior changes are detected, rather than processing the entire image frame. The active region extraction uses a simplified brightness-based method that requires minimal computation, and only the resulting regions of interest are subjected to full deep learning analysis. This partial application of the complex algorithm maintains detection effectiveness while reducing overall computational burden.
2Measurement precision
If a deep learning object detection algorithm is applied to detect driver behavior, then detection accuracy is improved, but processing speed decreases due to high computational requirements
Solution Approach 1:
The patent segments the image processing workflow into a fast preliminary stage (active region extraction using brightness comparison) and a detailed analysis stage (deep learning object detection). By dividing the task this way, the system achieves high processing speed in the first stage and maintains high detection accuracy in the second stage, as the deep learning algorithm processes only small extracted regions rather than full images.
Solution Approach 2:
The patent performs preliminary action by executing the active region extraction algorithm before applying the deep learning object detection algorithm. This preliminary step uses simple brightness-based active image production to identify and extract regions where driver behavior changes occur. By preparing and isolating these regions in advance, the system enables faster subsequent processing by the deep learning algorithm, improving overall processing speed while maintaining accuracy.
3Measurement precision
If full-resolution images are processed for driver behavior detection, then detection accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the full-resolution image processing into two phases: first, processing the entire image at full resolution using the lightweight active region extraction algorithm to identify behavior regions; second, processing only the extracted active regions using the deep learning algorithm. This segmentation maintains the ability to detect subtle behavior changes in the original high-resolution data while reducing the computational complexity of the resource-intensive deep learning processing by limiting it to small region-of-interest crops.
Data Source
AI summary
A driver behavior detection system using deep learning-based object classification includes: a frame inputting unit for receiving image frames; a downsampling unit for downsampling resolutions of a previous frame and a current frame of the image frames; an active image producing unit for utilizing brightness values by color of the downsampled previous and current frames to produce an active image; an active region extracting unit for applying a sliding window algorithm to the produced active image to extract an active region having the biggest window value among window values; and a behavior detecting unit for applying an object classification algorithm to the extracted active region to classify and detect a driver's behavior.