Deep learning in computer vision and sensor fusion
Learning outcomes: Attendees to this course will leave with a good sense of how deep learning can be used for a range of computer vision tasks and sensor fusion technology. In addition to the theoretical content, students will learn to design, train and debug neural networks in practice on real computer vision tasks and data.
Content: Application of computer vision and sensor fusion technology has drawn a lot of industrial and academic interest in recent years. They are widely used in many real applications such as autonomous vehicles, remote sensing, video surveillance and military. Deep Learning (DL) successfully applied to a wide range of computer vision tasks showing state-of-the-art performance. This course consists of two main parts.
The goal of the first part is to give a brief introduction of DL and then turning to introduce a most popular DL architecture (Convolutional Neural Networks) with a focus on learning common CNN models for main computer vision tasks: image classification, object localization, object detection, semantic segmentation and instance segmentation.
Second part gives an overview of sensor fusion techniques and modern sensors such as camera, radar and Lidar in the field of computer vision. Introduction to main DL-based techniques for image fusion, multi-source fusion and depth image prediction. Application of multi-senor fusion in autonomous driving and target recognition will be discussed.
Study methods: Lectures, practical project (individual and team work) and exam testing the understanding of essential concepts. The recorded lectures can be accessed through Moodle course page. There is only an online session after each two lectures for questions and assignments.
More information in University of Turku’s course page.
syväoppiminen tietokonenäkö neuroverkot luokittelu
Contact person for applications