Student of Natural Sciences, Engineering or Computer Science (f/m/x)

What to expect

Road marking extraction from aerial imagery is a key component in building and maintaining digital and HD maps. Accurate detection and vectorization of markings such as lane lines, arrows, crosswalks, and stop lines can significantly enhance applications in autonomous vehicle localization and navigation, HD map updating, and road condition assessment. Due to challenges in extracting fine details from aerial imagery, using information from other modalities, such as street-view imagery, could be beneficial. Developing robust AI models based on e.g., ViTs, generativeAI, diffusion, ... for this task is crucial, as reliable and up-to-date information in large-scale supports both traffic safety and intelligent transportation systems. This thesis will focus on designing and implementing a deep learning-based pipeline to efficiently and effectively extract road markings in vector format, ready for integration into digital mapping frameworks.

Your tasks

• Detecting road markings from Aerial imagery using Deep Neural Networks
• Exploring Street-View imagery for multimodal Deep Neural Network training

Your profile

• Ongoing Master’s studies in Computer Science, AI, Computer Vision, Geoinformatics, or a related discipline
• Programming experience in Python
• Experience in AI and Deep Learning approaches
• Experience with Pytorch or Tensorflow frameworks
• Good communication skills and proficiency in English (spoken and written)
• Basic knowledge of semantic segmentation or/and object detection