ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Download
Publications Copernicus
Download
Citation
Articles | Volume V-2-2020
ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci., V-2-2020, 917–924, 2020
https://doi.org/10.5194/isprs-annals-V-2-2020-917-2020
ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci., V-2-2020, 917–924, 2020
https://doi.org/10.5194/isprs-annals-V-2-2020-917-2020

  03 Aug 2020

03 Aug 2020

LEARNING WITH REAL-WORLD AND ARTIFICIAL DATA FOR IMPROVED VEHICLE DETECTION IN AERIAL IMAGERY

I. Weber1, J. Bongartz1, and R. Roscher2 I. Weber et al.
  • 1University of Applied Sciences Koblenz, AMLS, Remagen, Germany
  • 2Institute of Geodesy and Geoinformation, University of Bonn, Germany

Keywords: object detection, deep learning, data generation, multi-source learning

Abstract. Detecting objects in aerial images is an important task in different environmental and infrastructure-related applications. Deep learning object detectors like RetinaNet offer decent detection performance; however, they require a large amount of annotated training data. It is well known that the collection of annotated data is a time consuming and tedious task, which often cannot be performed sufficiently well for remote sensing tasks since the required data must cover a wide variety of scenes and objects. In this paper, we analyze the performance of such a network given a limited amount of training data and address the research question of whether artificially generated training data can be used to overcome the challenge of real-world data sets with a small amount of training data. For our experiments, we use the ISPRS 2D Semantic Labeling Contest Potsdam data set for vehicle detection, where we derive object-bounding boxes of vehicles suitable for our task. We generate artificial data based on vehicle blueprints and show that networks trained only on generated data may have a lower performance, but are still able to detect most of the vehicles found in the real data set. Moreover, we show that adding generated data to real-world data sets with a limited amount of training data, the performance can be increased significantly, and in some cases, almost reach baseline performance levels.