Automated evaluation of pig consciousness in the slaughterhouse: sensor development

AUTOMATED EVALUATION OF PIG CONSCIOUSNESS IN THE SLAUGHTERHOUSE: SENSOR DEVELOPMENT Angela Ramon-Perez , Sofia Gol , Josep Reixach , Jordi Sanchez , Xavier Manteca , Pol Llonch 1 2 2 3 1 1 Department of Animal and Food Science, School of Veterinary Science, Autonomous University of Barcelona, Building V, Travessera dels Turons, 08193 Cerdanyola del Vallès, Barcelona, Spain. 1 Selección Batallé S.A, Riudarenes, Girona, Spain 2 Nuna Solutions, IT & Industrial Automation for Meat Industry, Calafell, Catalonia, Spain 3 Caring beyond the farm gate Registered Charity in England no 1159690: Charitable Incorporated Organisation Humane Slaughter Association Animal welfare during slaughter is an ethical, legal, and social priority in pig production. Among the different phases of the slaughter process, stunning is one of the most critical points as it must induce a rapid and effective loss of consciousness. However, monitoring the state of consciousness in each pig is labourintensive and requires considerable human resources due to the high line speeds. Although a combination of several indicators is needed for assessing unconsciousness accurately, the corneal reflex is widely used since it is very sensitive and is one of the first indicators to reappear when the animal starts regaining consciousness. Technology and artificial intelligence are becoming promising tools for continuous, real-time, and individualised assessments in slaughterhouses, although commercial application remains a challenge. This ongoing study aims to develop and evaluate the feasibility of an automated sensor for monitoring pig consciousness in a commercial slaughterhouse by detecting the corneal reflex. The prototype sensor is a system integrating automatic blink detection using computer vision. It consists of a high-quality IP camera connected to an air blower directed towards the pigs’ cornea. A Raspberry Pi PLC handles digital I/O control, synchronising the activation of the air stimulus and the video capture. Video sequences are then processed by a GPU-enabled industrial PC running a neural network based on YOLO and OpenCV, which detects the presence or absence of the corneal reflex. The system provides real-time alerts via coloured light signals onsite and also sends notifications through a dedicated app, allowing for immediate corrective action if needed. The study is being conducted in a commercial slaughterhouse operating at a chain speed of 550 pigs per hour, using CO2 stunning. The trial has been performed on three days, collecting 1737 assessments of corneal reflex automatically to monitor the stunning effectiveness. From all the assessments, only five animals showed a positive corneal reflex. The low number of positive cases represents a limitation for this study, so further testing and data collection are needed to increase the sample size for algorithm development and later validation under commercial conditions. Despite this limitation, the current dataset is already contributing to the training of the neural network, and the system is designed to continuously improve its accuracy as new data are collected in real slaughterhouse conditions. This automated assessment approach could significantly enhance the objective and effective monitoring of animal welfare in slaughterhouses. 29 HSA Conference 2025: Animal Welfare Beyond the Farm Gate: Latest developments in humane slaughter and transport of animals