Geometry Projection Based 3D Object Detection

Unmanned aerial vehicles (UAV) have been deployed in many real-world applications such as payload delivery, agricultural laboring, and aerial photography. Based on GPS-based localization algorithms, these drones have achieved many advancements in industry automation. However, since common GPS receivers do not work indoors, an accurate computer vision (CV) technique is required to resolve the problem in indoor-UAV localization and navigation. In this research project, we aim to address the problem of 3D object detection, tracking and analysis of indoor UAV by deploying a deep-learning-based CV algorithm. The success of this project will support the development of indoor UAV navigation and task-related operation. Also, the study in this field will expand the application in computer vision and boost the full automation in warehouse inventory management.

Faculty Supervisor:

Igor Gilitschenski;Babak Taati;Steven Waslander

Student:

Partner:

SOTI Inc

Discipline:

Computer science

Sector:

Information and cultural industries; Professional, scientific and technical services

University:

University of Toronto

Program: