Development of a semi-automated ground truth annotation system for intelligent transport system applications

• Main Author: Rivera, Maverick C.
• Format: text
• Language: English
• Published: Animo Repository 2021
• Subjects: Intelligent transportation systems; Computer vision; Image processing—Digital techniques; Electrical and Computer Engineering
• Online Access: https://animorepository.dlsu.edu.ph/etdm_ece/4; https://animorepository.dlsu.edu.ph/cgi/viewcontent.cgi?article=1003&context=etdm_ece
• Institution: De La Salle University

Image and video data annotation is an effort and time-consuming process. However, it is necessary for training neural network models for computer vision applications such as in Intelligent Transport Systems. Thus, this research explores ways in order to reduce the time and effort necessary to establish ground truth annotations for videos and images specifically for ITS applications. The approach of this research is by using existing computer vision and machine learning tools and methodologies to aid the annotation process. Mainly, the research explored the use of trained models to create the initial bounding boxes in the video or image being annotated. The generated bounding boxes are then tracked using Siamese-based trackers with the aid of a human annotator which modifies the result as necessary.