Multi-Task Learning of Height and Semantics from Aerial Images
This work addresses the need for efficient land surface analysis from aerial imagery, offering a multi-task approach that enhances accuracy for applications like land use mapping and elevation modeling, though it is incremental in nature.
The authors tackled the problem of jointly learning semantics and local height from aerial images using a multi-task neural network framework, achieving improved performance on both tasks on the 2018 Data Fusion Contest dataset and providing uncertainty maps for prediction assessment.
Aerial or satellite imagery is a great source for land surface analysis, which might yield land use maps or elevation models. In this investigation, we present a neural network framework for learning semantics and local height together. We show how this joint multi-task learning benefits to each task on the large dataset of the 2018 Data Fusion Contest. Moreover, our framework also yields an uncertainty map which allows assessing the prediction of the model. Code is available at https://github.com/marcelampc/mtl_aerial_images .