Short description

There are several reasons why one would want to monitor the appearance of birds in a certain area. First, there is a big community of ornithologists who want to follow the appearance birds in the environment. Some birds only pass through a certain area when migrating and are therefore only hard to track. In the course of investigations of the environmental impact of large building projects the area has to be surveyed for endangered birds. Especially large birds are at risk of being killed by wind power stations, therefore sometimes those are shut down when birds are approaching. An automatic scanning would help, though this is already done by using radar systems. Since nobody can do this 24/7 automatic detection and recording of the passing birds will be helpful.

• Modules/Topics:
  • Acoustic Processing
    • Microphone Array
      • Source Localization and/or Beamforming
    • Noise Reduction
    • Automatic Bird Species Classification
  • GPS Sensor for automatic position and time reference
  • Optical Processing:
    • Camera that records the sky
      • Movement detection
    • Bird flock detection and recording
  • Flight path tracking
  • PV-Module and battery for off-grid operation
  • Weather resistant housing
    • Microphones and camera need to withstand rain and wind.

This projects needs to be a team project, where every member focuses on one topic. At the beginning of-the-shelve components can be used. They will be improved, bepending on your focus.

References:

• Irina Tolkova and Holger Klinck: Source separation with an acoustic vector sensor for terrestrial bioacoustics, The Journal of the Acoustical Society of America , Vol. 152, No. 2, p. 1123, 2022, DOI: https://doi.org/10.1121/10.0013505
• Gerald Bota, et al.: Hearing to the Unseen: AudioMoth and BirdNET as a Cheap and Easy Method for Monitoring Cryptic Bird Species, Sensors , Vol. 23, No. 16, p. 7176, 2019, DOI: https://doi.org/10.3390/s23167176
• Stefan Kahl, et al.: BirdNET: A deep learning solution for avian diversity monitoring , in Ecological Informatics , Vol. 61, p. 101236, 2021, DOI: https://doi.org/10.1016/j.ecoinf.2021.101236
• Becky E. Heath, et al.: MAARU: Multichannel Acoustic Autonomous Recording Unit for spatial ecosystem monitoring., bioRxiv preprint, DOI: https://doi.org/10.1101/2024.01.23.576628, 2024

Tasks (depending on specific project):

• Review of literature and potential algorithms
• Implementation of algorithms in Python and/or Julia
• Design hardware
• Choose hardware platform
• Evaluate the chosen approach(es)
• Implementation on power efficient hardware (e.g. ARM Cortex-M7)
• Documentation

Your Profile/Prerequisites

• Motivation and interest in the topic
• Background depending of focus of project e.g. in Audio Signal Processing, Machine Learning
• Experience with coding e.g. (Matlab, Python or Julia, C, rust)
• Only in teams of 2-4 students

Contact:

Martin Hagmüller (hagmueller@tugraz.at or 0316/873 4377)