The joint research project ‘Sound based parameters as bio-acoustic indicators for the animal welfare of fattening turkeys’ engages with the automatic analysis of the behavior and health of fattening turkeys in a conventional housing system and contributes towards an improvement of appropriate animal welfare, housing and health.

With the help of machine learning, this project examines if animal related sounds or other noises are suitable bioacoustics indicators that give information about the current behavior and general well-being of the flock. For this purpose, optic and acoustic recordings as well as data about the condition of the animals are simultaneously generated for the whole fattening period in the experimental barn. The acoustic data are recorded and pre-processed with microphones and furthermore used as the Input for Deep-Learning AI-models. The data of the camera specially serve for the evaluation of training data, whereas the analysis can partly be done automatically. The aim is to examine two different kinds of events: On the one hand, the general condition of the flock will continuously be assessed to detect discrepancies from the expected values in an early stage (e.g. broadening infections). In this case, temporally slow variances of the ambient noises in the barn are expected. On the other hand, it is the aim to detect antagonistic and potentially harmful behavior, as early as possible. Bio-acoustic indicators in this context are temporally rather transient sounds with rapid local change, for which acoustic beamforming or acoustic camera techniques are applied.

Project management:  

Dr. Helen Schomburg
Dr. Thomas Bartels

Project staff: 

M.Sc. Jessica Raabe

Cooperation partner:

Prof. Dr.-Ing. Martin Streitenberger
Prof. Dr. Ing. Kai Homeyer

Hochschule Hannover
Fakultät I - Elektro- und Informationstechnik
Ricklinger Stadtweg 120
D-30459 Hannover