Lars Pandikow
About Lars Pandikow
Lars Pandikow is a Machine Learning Engineer currently working at Parallel Domain in Vancouver, Canada. He has a background in Visual Computing and Computer Science from Technische Universität Darmstadt and has contributed to advancements in cyclist detection using machine learning techniques.
Current Role at Parallel Domain
Lars Pandikow holds the position of Machine Learning Engineer at Parallel Domain, where he has been employed since 2019. His work focuses on developing and enhancing machine learning models, particularly in the area of computer vision. He has contributed to significant projects that leverage advanced algorithms to improve object detection capabilities.
Previous Experience at Fraunhofer IGD
Prior to his current role, Lars Pandikow worked at Fraunhofer IGD in the Darmstadt Area, Germany. He served as a Software Developer for five months in 2019 and as a Student Research Assistant from 2014 to 2019. During his tenure, he gained valuable experience in software development and research, contributing to various projects in visual computing.
Education and Academic Background
Lars Pandikow studied at Technische Universität Darmstadt, where he earned a Master of Science (MS) in Visual Computing. He also obtained a Bachelor of Science (BS) in Computer Science from the same institution. His academic background provided him with a strong foundation in computational techniques and visual analysis.
Research Contributions and Projects
Lars Pandikow has made notable contributions to the field of machine learning, particularly in cyclist detection. He was involved in significant improvements in 2D bounding box detection for cyclists using the YOLOv3 model. His research efforts led to double-digit improvements in average precision for cyclist detection and included the publication of a blog post on enhancing cyclist detection through synthetic data.
Dataset Enhancement Initiatives
In his research, Lars Pandikow worked on supplementing the KITTI and nuImages datasets with synthetic data. This initiative aimed to enhance the quality and quantity of data available for training machine learning models, thereby improving the performance of detection algorithms in real-world applications.