Tran, Albert: Artificial Intelligence-Aided Railroad Trespassing Data Analytics: Methodology and A Case Study

View video presentation

Title:  Artificial Intelligence-Aided Railroad Trespassing Data Analytics: Methodology and A Case Study

Name:  Albert Tran

Major:  Electrical and Computer Engineering

School affiliation:  School of Engineering

Programs:  Aresty – RA Program

Other contributors:    Zhipeng Zhang, Jinxuan Xu, and Xiang Lu

Abstract:  Using video data and artificial intelligence, real time detection of railroad trespassing can be obtained. Of all the
rail-related fatalities, trespassing is one of the top causes and little progress has been done to stop the prevention of
such accidents. While there are many surveillance cameras to witness such accidents, the cost of employing people to
continuously monitor the video still introduces drawbacks.  However, using the real time video from the cameras and
computer vision, algorithms can take the role of detecting trespassing at the railways, eliminating the need of human
monitoring.  A computer vision algorithm, You-Only-Look-Once (YOLO),  that can detect objects in real time, is used to create a framework that the railroad industry can use to change the infrastructure of the locations where trespassing occurs in
order to safeguard from accidents. From that datagenerated from this framework, data such as: the time of
day, weather, type of trespassing, and location on railway of trespassing can be outputted on detection and stored in a
database for easy processing.