Carl Schwedes

• Focused to work and the development of algorithms for the purpose of semantic ground-plane segmentation under the use of deep learning application (Master thesis). (Tensorflow, CUDA, Docker, team organization via Scrum, Jira, Bitbucket)

• Undertaking research to develop and enhance lane detection and tracking algorithms based on video-sensors, image analysis, edge detection algorithms, kalman-filter tracking, real-time operating systems (PREEMPT-RT), CAN-Protocol, PID-Controller and Robot Operating System (ROS). • Recoded and redesigned OpenCV image analysing algorithms to achieve further improvements on efficiency.

• Master thesis: Semantic Lane Segmentation of LiDAR Point Clouds using Convolutional Neural Networks. Keywords: environmental perception, LiDAR, deep learning, tensorflow, nuScenes, focal loss, range-image projection • Specialized in deep learning targeting computer vision applications

• Included coding with LWJGL and OpenGL. The project comprised development of a molecular viewer application with the purpose to visualize molecular mechanisms of antibiotic resistance related to modern penicillin drugs (part of the BACK FROM THE DEAD: Demystifying Antibiotics special exhibition).

• The project included an article, as well as the conception and design of 3d visualization content for educational purposes to help and support the understanding of historical production processes. The project was part of the: ”BACK FROM THE DEAD: Demystifying Antibiotics” special exhibition.