GPU Embedded Vision for Roadway Asset Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Manual tracking and monitoring of thousands of assets along roadways, such as streetlights, signboards, and potholes, is tedious and impractical due to the high manual intervention required, necessitating a system that can detect, track, and monitor assets with minimal hardware and efficiently.
Innovation Solution
A GPU-based embedded computer vision system utilizing YOLOACT and YoloV4 algorithms, combined with a tracking, mapping, and localizing module, to identify and track assets by comparing video data with master videos, assigning trackers, and mapping geo-coordinates, which operates with minimal computational power and can detect missing or damaged assets in a single frame.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If manual tracking and monitoring of thousands of assets along roadways is performed, then asset condition tracking is achieved, but high manual intervention and effort are required making it tedious and impractical
Solution Approach 1:
The patent replaces manual mechanical tracking methods with an automated computer vision system using GPU-based deep learning algorithms (YOLO). The system captures images via camera, processes them through neural networks for object detection and classification, and automatically tracks assets without human intervention, thereby substituting mechanical/manual operations with an intelligent automated system.
2Measurement precision
If complex object detection algorithms are used to achieve high accuracy in asset detection, then detection precision is improved, but computational power requirements increase
Solution Approach 1:
The patent employs YOLO (You Only Look Once) algorithm which changes the computational approach by performing single-shot detection instead of multiple sequential processing steps. This parameter change in the detection methodology achieves high accuracy while reducing computational power consumption compared to traditional multi-stage object detection algorithms.
3Productivity
If traditional object detection systems are deployed to detect and track roadway assets, then detection capability is achieved, but hardware requirements and system cost increase
Solution Approach 1:
The patent creates a multi-functional integrated system where a single GPU-based embedded platform performs camera capture, image processing, object detection, classification, tracking, and geo-coordinate mapping. This universal system handles multiple asset types (streetlights, signboards, potholes, etc.) and multiple functions within one hardware platform, reducing overall device complexity and cost compared to separate specialized systems.
Data Source
AI summary
The present invention relates to system and method that detects objects and assigns trackers to each object and maps it to detections to its corresponding Geo-Coordinates. Based on the number of detections, it assigns new trackers and counts the objects identified using optimal and efficient computation. The present system requires just one detection and its location to create a tracker accurately. Also, if an object is missing in any given location, the present system identifies the Geo-Coordinates and shows the image along with its details.