Dr Yuri Shendryk1, Yannik Rist1, Dr Catherine Ticehurst2, Dr Peter Thorburn1
1CSIRO, Brisbane, Australia,
2CSIRO, Canberra, Australia
Motivated by the increasing availability of high-resolution satellite imagery, we developed deep learning models able to efficiently and accurately classify the atmospheric conditions and dominant classes of land cover/land use in commercial PlanetScope imagery acquired over the Amazon rainforest. In specific, we trained deep convolutional neural network (CNN) to perform multi-label scene classification of high-resolution (<10 m) satellite imagery. We also discuss the challenges and opportunities in training ensemble CNN models for multi-label scene classification. Our best performing model achieved an F-beta score of 0.91, which was only 2% short of the top performing model in the Understanding the Amazon from Space Kaggle competition. Finally, we investigate the transferability of our PlanetScope-trained models to freely available Sentinel-2 imagery acquired over the wet tropics of Australia. The models trained on PlanetScope imagery performed well when applied to Sentinel-2 imagery with F-beta score of 0.79. Similarly, the models trained on Sentinel-2 imagery achieved F-beta score of 0.80 when applied to PlanetScope imagery. This suggests that our CNN models are suitable for classifying the atmospheric conditions and dominant classes of land cover/land use in satellite imagery of similar resolution to that of PlanetScope and Sentinel-2 (i.e. <10 m).
Biography:
Dr Yuri Shendryk is a Postdoctoral Fellow at CSIRO specializing in developing algorithms to process terabytes of satellite and airborne data. After earning a master’s degree in geophysics, he spent more than three years working and studying geospatial engineering in Sweden and Germany. In 2017 he received a PhD from UNSW, and his current research in CSIRO is centred around the integration of remote sensing and machine learning for forest health monitoring and precision agriculture.