Active Learning for UAV-based Semantic Mapping

2019/08/29 by Hermann Blum, Blum, Hermann, Silvan Rohrbach +7
Computer Science · Engineering · #FOS: Computer and information sciences #Machine Learning and Algorithms #Robotic Path Planning Algorithms #Robotics (cs.RO) #Robotics and Sensor-Based Localization

paper · pdf · doi:10.48550/arxiv.1908.11157

openalex publication_date 2019/08/29 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Unmanned aerial vehicles combined with computer vision systems, such as convolutional neural networks, offer a flexible and affordable solution for terrain monitoring, mapping, and detection tasks. However, a key challenge remains the collection and annotation of training data for the given sensors, application, and mission. We introduce an informative path planning system that incorporates novelty estimation into its objective function, based on research for uncertainty estimation in deep learning. The system is designed for data collection to reduce both the number of flights and of annotated images. We evaluate the approach on real world terrain mapping data and show significantly smaller collected training dataset compared to standard lawnmower data collection techniques.

Citations

• Planning and acting in partially observable stochastic domains
• What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision?

Related

• Informative Path Planning for Active Learning in Aerial Semantic Mapping
• An Informative Path Planning Framework for Active Learning in UAV-based Semantic Mapping
• Adaptive Path Planning for UAV-based Multi-Resolution Semantic Segmentation
• Adaptive Path Planning for UAVs for Multi-Resolution Semantic Segmentation
• An informative path planning framework for UAV-based terrain monitoring