Vers une détection automatisée des comportements délétères des porcs en élevage
In: Journées de la Recherche Porcine en France ; 51. Journées de la Recherche Porcine ; https://hal.science/hal-02097194 ; 51. Journées de la Recherche Porcine, Feb 2019, Paris, 2019
Online
Konferenz
Zugriff:
Aggressive behaviours harm the health and welfare of pigs, and in turn negatively affect the productivity of farms. INRA and CEA are collaborating in the European project PIGWATCH (ERANET Anihwa) to develop an automatic system to detect harmful behaviours such as fighting. The system is based on sensors and algorithms of artificial intelligence. In this study, the system (developed by CEA-LETI) included a triaxial accelerometer and was fixed on the ears of the pigs. It was connected to an Android application, enabling data to be transferred to a smartphone via Bluetooth. Twelve pigs were fitted with the system. Their activity was recorded regularly with a video camera for two months. Pig behaviours, especially fighting, were scored from about 24 hours of recordings. Signals from the sensors were marked as a function of the behaviours observed. Mathematical characteristics of the data that differed according to the behaviours displayed at the time were extracted. We then developed an algorithm using these characteristics to detect behaviours automatically. Currently, the algorithm can detect aggressive behaviours with a sensitivity of 41% and a specificity of 87%. An improvement is expected by increasing the data basis. Preliminary analyses of variations during the day in behavioural activities obtained from predictions of the system show typical variations in agreement with preceding studies, with high levels of activity in the early morning and afternoon. These results are encouraging for the potential of the system to automatically record behaviours on farms. ; Les comportements agressifs sont délétères et affectent la santé et le bien-être des porcs ainsi que la productivité des élevages. Dans le cadre du projet européen PIGWATCH (ERANET Anihwa), l’INRA et le CEA travaillent au développement d’une technique automatisée, basée sur des capteurs et des algorithmes d’intelligence artificielle, pour détecter les comportements délétères de type bagarres. Le CEA-LETI a développé un dispositif porté à l’oreille ...
Titel: |
Vers une détection automatisée des comportements délétères des porcs en élevage
|
---|---|
Autor/in / Beteiligte Person: | Prunier, Armelle ; Tallet, Céline ; Lagarrigues, Guillaume ; Sakri, Oumayma ; Vidal, Audrey ; Coulon Lauture, Françoise ; Brulais, Sébastien ; Godin, Christelle ; Labyt, Etienne ; Physiologie, Environnement et Génétique pour l'Animal et les Systèmes d'Elevage Rennes (PEGASE) ; Institut National de la Recherche Agronomique (INRA)-AGROCAMPUS, OUEST ; ProbaYes ; LETI Department Systems and Solutions Integration ; Commissariat à l'Energie Atomique et aux Energies Alternatives (CEA) |
Link: | |
Zeitschrift: | Journées de la Recherche Porcine en France ; 51. Journées de la Recherche Porcine ; https://hal.science/hal-02097194 ; 51. Journées de la Recherche Porcine, Feb 2019, Paris, 2019 |
Veröffentlichung: | HAL CCSD ; IFIP - Institut du Porc, 2019 |
Medientyp: | Konferenz |
ISBN: | 978-2-85969-241-4 (print) ; 2-85969-241-X (print) |
Schlagwort: |
|
Sonstiges: |
|