A highly efficient and elective attribute learning framework for road graph from aerial imagery and GPS

Road attributes play a pivotal role in digital maps, providing critical information for various routing and planning applications that aim to create a safe and efficient traffic environment. While some road attributes are available in existing map data such as OpenStreetMap [3], these sources may not cover all regions, meet highquality standards, or include specific attributes required for specialized applications using these. To address these challenges, we propose a novel framework that leverages multi-task deep learning to learn road attributes from remote sensing imagery and GPS data. Our approach treats the task as a multi-task learning problem and incorporates convolutional and graph neural networks into an end-to-end learning framework. This enables efficient prediction of multiple road attributes for a set of input roads. To evaluate our system, we collect annotations and develop our model using public map sources. Our results demonstrate promising performance in predicting road type, road median, lane number, road directionality, and width in meters. By exploring different road attributes
compared to previous works, our efforts open up new possibilities for novel applications in this domain. Overall, our research contributes to advancing the understanding and prediction of road attributes, enhancing the quality and completeness of digital maps, and enabling the development of innovative solutions for various applications.