Vehicle Pedestrian Detection Using Feature Point Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Advanced driver assistance systems (ADAS) and autonomous driving systems face challenges in accurately detecting pedestrians and bicycle riders who are distracted by music or images, leading to a high risk of accidents due to their lack of attention to their surroundings.
Innovation Solution
A method and apparatus that use image detection to identify users of devices such as headphones or mobile devices, extracting static and dynamic feature points to determine if a pedestrian or bicycle rider is distracted, and then inform them and vehicle occupants of an approaching vehicle through auditory, visual, or tactile means.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If ADAS uses conventional sensors to detect pedestrians, then detection coverage is achieved, but detection accuracy is insufficient for distracted pedestrians
Solution Approach 1:
The system segments the detection process into multiple stages: initial object detection, device usage detection through feature point extraction, and distraction determination. This segmentation allows the system to focus computational resources on identifying distracted pedestrians rather than processing all detected objects uniformly, thereby improving detection accuracy while managing system complexity.
Solution Approach 2:
The system transitions from conventional 2D image detection to 3D spatial analysis by extracting feature points in three-dimensional space and calculating distances between the vehicle and pedestrians. This dimensional enhancement improves detection accuracy by providing depth information and spatial relationships that conventional 2D sensors cannot capture.
2Reliability
If the system alerts all detected pedestrians, then safety coverage is maximized, but false alarms increase for non-distracted pedestrians
Solution Approach 1:
The system implements feedback mechanisms by continuously monitoring pedestrian behavior through feature point tracking and analyzing changes in pedestrian state over time. This feedback loop allows the system to distinguish between distracted and non-distracted pedestrians based on behavioral patterns, improving alert reliability while reducing false alarms.
Solution Approach 2:
The system replaces conventional mechanical sensor-based detection with vision-based optical detection using cameras and image processing. This substitution enables the system to analyze pedestrian behavior, device usage, and distraction states through visual information, thereby improving the ability to distinguish distracted pedestrians without increasing false alarms.
3Measurement precision
If the system uses multiple sensing methods to detect distracted pedestrians, then detection accuracy improves, but processing time increases
Solution Approach 1:
The system performs preliminary actions by pre-defining feature points and detection criteria before actual pedestrian detection occurs. Feature points such as ear positions, hand positions, and device locations are predetermined, allowing the system to quickly match detected objects against these pre-established patterns. This preliminary preparation significantly reduces processing time while maintaining high detection accuracy.
Solution Approach 2:
The system dynamically adjusts detection parameters such as feature point thresholds, distance thresholds, and alert triggers based on environmental conditions and pedestrian behavior patterns. By changing these parameters adaptively, the system optimizes the balance between detection accuracy and processing time, avoiding unnecessary computations for clear-cut cases while maintaining high accuracy for ambiguous situations.
Data Source
AI summary
Disclosed is a method and apparatus of detecting an object approaching a vehicle, the method includes a user from the image, extracting feature points associated with the device or a portion of the user's body from the image of the user, detecting the object using a device based on the feature point, and issuing an alert of the approaching object.