Open YPEC 2022

ON04 – Object Detection using video analytics (for railway assets)

MTR Data Studio has developed a project about railway object detection. The platform adopted AI, IoT and cloud technologies, with ease to strengthen Hong Kong status as an international asset management center and in line with the National 14th Five-Year Plan. The team has successfully developed the monitoring system to detect railway assets by using video analytics with the consideration of financial and technical feasibility. In this project, we develop several object detection based models, a computer vision technique that identifies and classifies a particular object in a particular setting. The main goal of object detection is to scan digital images or real-life scenarios to locate instances of every object, separate them, and analyze their necessary features for real-time predictions. we are managed to classified images and performed perfectly in 99% and error image can be filtered out. The object detection also performed feasibly well in more than 90% accuracy for most of the normal class, and will be further developed with more labelled data, advanced model and machine learning pipeline. The developed model powers a system to visually analyze each object and determines their real-life application, just like humans do. It automates the asset management process, reduce manpower and the human error for checking.