In-Vehicle Moving Vehicle Identification via Instance Segmentation and IoU
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for identifying moving vehicles through image processing, such as semantic segmentation, struggle to accurately distinguish between stationary and moving vehicles of the same category, leading to incorrect identifications and reduced accuracy in traffic safety assessments.
Innovation Solution
An in-vehicle device employs an instance segmentation algorithm combined with a monocular depth estimation residual convolutional neural network to capture and process images of a target vehicle at different times, generating mask areas and calculating the Intersection over Union (IoU) to determine the dynamic class object mask area, thereby distinguishing moving vehicles and ensuring accurate identification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If semantic segmentation is used to identify objects in images, then objects can be classified into categories, but different objects in the same category cannot be distinguished, leading to wrong identification of stationary vehicles as moving
Solution Approach 1:
The patent applies instance segmentation to divide the image into multiple segments, where each segment corresponds to a specific object instance. This allows differentiation between multiple vehicles of the same category by creating separate mask areas for each detected vehicle, thereby resolving the limitation of semantic segmentation that cannot distinguish individual objects within the same category.
2Measurement precision
If instance segmentation algorithm is used to capture images at different times and calculate IoU, then moving vehicles can be accurately distinguished, but the processing complexity increases
Solution Approach 1:
The patent performs preliminary actions by capturing images at multiple time points and generating mask areas for each detected vehicle before calculating IoU. This preliminary segmentation and temporal sampling prepare the data in advance, making the subsequent motion detection through IoU calculation more efficient and accurate, rather than attempting complex real-time motion analysis.
Solution Approach 2:
The patent introduces IoU (Intersection over Union) as an intermediary metric to compare mask areas from different time points. This intermediary calculation provides a quantitative measure of object displacement, enabling accurate motion detection through a simple comparison operation rather than complex direct coordinate analysis.
Data Source
AI summary
A method for identifying road vehicles or other objects which are in motion against those which are not moving applied in an in-vehicle device of an assisted vehicle which is being driven shoots a first image of a target vehicle and a second later image of the target vehicle, determines a first mask area of the target vehicle from the first image, and determines a second mask of the target vehicle from the second image based on an instance segmentation algorithm. An Intersection over Union (IoU) is calculated between the first mask area and the second mask area and a determination made as to whether a dynamic class object mask area of the target vehicle according to the IoU should be generated. A dynamic class object mask area of the target vehicle is generated when the target vehicle is found to be a moving vehicle.