Main Content

Deep Learning for Environmental Monitoring

Research in the field of artificial intelligence (AI) has made great progress in recent years. Deep learning in particular, i.e. the use of deep artificial neural networks, is currently experiencing a major upswing. This is made possible by massive increases in the computing capacity of modern graphics cards, by the availability of data sets with millions of training examples and, last but not least, by new technologies that make the learning of deep network architectures possible in the first place.

In the course of this development, more and more research areas are being opened up as application areas for Deep Learning-based learning methods. One particularly important area of application for the preservation of our environment is nature and environmental monitoring, i.e. the automated observation and study of nature and the environment. Here, data recorded by cameras, microphones or other sensors can be analysed with the help of artificial neural networks to recognise patterns (e.g. determination of an animal species) or to make predictions about future events (e.g. water shortage).

In this seminar, current Deep Learning methods for the investigation of nature and the environment will be presented and discussed. In addition to a basic understanding of Deep Learning, the aim is to become more familiar with individual methods for processing specific types of data (audio, images, videos, sensors). Possible topics include:

• Animal recognition in videos from photo traps
• Bird call recognition
• Monitoring forest fires using satellite and drone imagery
• Weather and disaster forecasting using sensor data

The participants should be enabled to gain targeted knowledge and experience in the field of Deep Learning with the specific application area of environmental investigation according to their own interests and focal points. This can then be built upon in a targeted and flexible manner in subsequent events.