0 avis
Development of automated sound indicator measurements for the early detection of respiratory disorders in poultry: the case of Infectious Bronchitis
Archive ouverte
Edité par CCSD -
International audience. The vocalizations produced by poultry provide information on their health and welfare. This information, complementary to that of the breeder, can allow early detection of a pathology. This technology would allow the farmer to be warned when alert criteria are exceeded so that appropriate corrective measures can be taken immediately, thereby limiting the aggravation of the problem. The objective of this work is to evaluate the feasibility to automatically detect symptoms of Infectious Bronchitis (sneezing and rales) in poultry. A second objective is to identify one or more acoustic indicators for early detection of the establishment of this virus. To achieve those objectives, soundtracks were collected under controlled experimental conditions from groups of ROSS 308 broilers infected (trial; n=30) and uninfected (control;n=30) with infectious bronchitis (avian coronavirus). A first stage of listening by a group of experts allowed the detection and isolation of rales and sneezes, thus those symptoms were have been characterized using acoustic descriptors. An algorithm was then developed to automatically and specifically detect and isolate these symptoms on the trial group. The second step was to differentiate the control group from the trial group based on the temporal evolution of the relative noise level. Three days post-inoculation, a difference of 3 decibels was observed at night between the groups (+3 dB for the infected group; sound intensity multiplied by 2). The algorithm developed allowed us to detect sneezing with 95% sensitivity. This work highlights the value of acoustic analysis for the early detection of pathologies in broilers. It should be continued with trials in field conditions.
Consulter en ligne
Suggestions
Du même auteur
