Speakers

View Full Calendar

Computer Vision-Based Vehicle Classification Framework

Event Type
Seminar/Symposium
Sponsor
New Frontiers Initiative and the Blue Waters Project
Date
Oct 12, 2021   10:00 - 11:00 am  
Speaker
Ramez Hajj, Department of Civil and Environmental Engineering, University of Illinois
Registration
Registration
Contact
New Frontiers Initiative
E-Mail
newfrontiers@illinois.edu
Phone
217-333-1163
Views
6
Originating Calendar
New Frontiers Initiative

Abstract: In the last few years, deep learning has, in conjunction with computer vision techniques, rapidly advanced to allow the classification of objects accurately and efficiently. With the increasing use of unmanned aerial vehicles (UAVs) and the availability of high-resolution satellite data, it is more important now than ever before to have effective algorithms for determining what objects are present. Vehicle classification and counting are important problems for intelligence applications. However, strong vehicle classification algorithms which can classify individual vehicles using satellite or UAV data have yet to be developed for use by intelligence agencies. This project seeks to develop vehicle classification and detection algorithms by leveraging the latest technologies in computer vision and deep learning. Aerial image data from labeling to validation are parameterized and monitored to feed into neural networks. The new framework lays focus on Machine Learning (ML) data pipelines, to improve the quality of production ML on the operationalization of models for accurate vehicle classification and counting. Transfer learning is used on different neural network architectures such as CenterNet, YOLOv4, and YOLOv5, which are all single stage object detection algorithms. The implemented  algorithms are subjected to hyperparameter tuning to obtain the best performing classification model or combination thereof.

link for robots only