Labeling of cough data from pigs for on-line disease monitoring by sound analysis

More objective and automated detection of respiratory diseases in pig houses should be possible by on-line sound analysis of cough monitoring. To develop automatic algorithms for pig cough recognition, experiments and well-labeled cough data are needed. The objectives of this article are: (1) to give a short overview of the attained results in cough recognition, and (2) to define a methodology to label the cough data in a pig house. Human observers labeled coughs by audiovisual observation in a laboratory test installation with ten pigs during periods ranging from two days to two weeks. Simultaneously, sound registration was done with audio equipment. The sound registrations were listened to by another observer to compare the number of coughs in the registration with the number labeled during the experiment. It was found that there were under estimations of up to 94% in the number of coughs. The underestimation in the number of coughs could be reduced to 10% when the observer used an additional labeling sound signal on the scene each time coughing was observed. In addition, differences were found between two independent observers scoring pig's coughs in an audiovisual manner on the scene. For future research, we suggest an investigation of how an observer using software labeling could improve the labeling results.