Computer Vision Based Real-Time Pixel-Level Railroad Track Components Detection System

Description:

Reference #: 01467

The University of South Carolina is offering licensing opportunities for Computer vision based real-time pixel-level railroad track components detection system

Background:

Rail inspection is the practice of examining railroads for flaws that could lead to catastrophic failures. This process takes a lot of time and is typically done using either visual inspection in person or with cameras. Current approaches are two steps, which separate inspection video collection and data analysis.

Invention Description:

The proposed invention is a real-time segmentation system using a deep learning algorithm and a camera to perform real-time component detection and recognition for a railroad track. This system changes railroad inspection from a two-step approach (first record inspection video and then bring the video back to analyze the video) into a one-step approach, analyzing the inspection video as it was recorded.

Potential Applications:

This system can be used for real-time railroad track inspection for detecting any missing or broken component along the track

Advantages and Benefits:

Current approaches are two steps, which separate inspection video collection and data analysis. All the collected videos need to be brought back from the field to process and obtain useful information. Our system combines the two steps together. The algorithm we developed can process the video as it was recorded. The field condition will be accessed immediately as it was inspection. No need to bring the videos back to postprocessing, which both increase the productivity and lower the inspection expenses.

Patent Information:
Title App Type Country Serial No. Patent No. File Date Issued Date Expire Date Patent Status
Computer Vision Based Real-Time Pixel-Level Railroad Track Components Detection System Utility United States 17/330,591   5/26/2021     Filed
For Information, Contact:
Technology Commercialization
University of South Carolina
technology@sc.edu
Inventors:
Yu Qian
Feng Guo
Keywords:
convolutional neural network
deep learning
image analysis
missing or broken component
railroad track inspection
real-time process
© 2024. All Rights Reserved. Powered by Inteum