2 results
Automatic classification of measures of lying to assess the lameness of broilers
- A Aydin, C Bahr, D Berckmans
-
- Journal:
- Animal Welfare / Volume 24 / Issue 3 / August 2015
- Published online by Cambridge University Press:
- 03 January 2023, pp. 335-343
-
- Article
- Export citation
-
Leg disorders are a major cause of poor welfare in broilers. Previous studies have shown that at slaughter age at least 90% of chickens experienced some degree of gait problems and approximately 30% were seriously lame. In this study, a new and non-invasive technique was developed to automatically assess the lameness of the birds. For this purpose, video surveillance images of broilers with five different pre-defined gait scores were recorded as they walked along a test corridor. Afterwards, the image-processing algorithm was applied to detect the number of lying events (NOL) and latency to lie down (LTL) of broiler chickens. Then, the results of the algorithm were compared with visually assessed manual labelling data (reference method) and the relation between these measures and lameness was investigated. Eighty-three percent of NOL were correctly classified by the automatic monitoring system when compared to manual labelling using a data set collected from 250 broiler chickens. The results also showed a positive significant correlation between NOL and gait score and a significant negative correlation between LTL and gait-score level of broilers. Since strong correlations were found, on the one hand, between two measures and gait-score level of broiler chickens and, on the other, between the results of algorithm and manual labelling, the results suggest this automatic monitoring system may have the potential to be used as a tool for assessing lameness of broiler chickens.
Light-based monitoring devices to assess range use by laying hens
- S. Buijs, C.J. Nicol, F. Booth, G. Richards, J.F. Tarlton
-
- Article
- Export citation
-
Access to an outdoor range has many potential benefits for laying hens but range use can be poor due to factors only partly understood. Techniques to monitor individual range use within commercial flocks are crucial to increase our understanding of these factors. Direct observation of individual range use is difficult and time-consuming, and automatic monitoring currently relies on equipment that is difficult to use in an on-farm setting without itself influencing range use. We evaluated the performance of a novel small, light and readily portable light-based monitoring system by validating its output against direct observations. Six commercial houses (2000 hens/house) and their adjacent ranges were used, three of which were equipped with more structures on the range than the others (to determine whether cover would influence monitoring accuracy). In each house, 14 hens were equipped with light monitoring devices for 5 discrete monitoring cycles of 7 to 8 consecutive days (at 20, 26, 32, 36 and 41 weeks of age). Light levels were determined each minute: if the reading on the hen-mounted device exceeded indoor light levels, the hen was classified as outside. Focal hens were observed directly for 5 min/hen per week. Accuracy (% of samples where monitoring and direct observations were in agreement) was high both for ranges with more and with fewer structures, although slightly better for the latter (92% v. 96% ± 1 SEM, F1,19 = 5.2, P = 0.034). Furthermore, accuracy increased over time (89%, 94%, 95%, 98% ± 1 SEM for observations at 26, 32, 36 and 41 weeks, respectively, F3,19 = 3.2, P = 0.047), probably due to progressively reduced indoor light levels resulting from partial closing of ventilation openings to sustain indoor temperature. Light-based monitoring was sufficiently accurate to indicate a tendency for a greater percentage of monitored time spent outside when more range structures were provided (more: 67%, fewer: 56%, SEM: 4, $\chi_1^2 = 2.9$, P = 0.089). Furthermore, clear and relatively consistent individual differences were detected. Individuals that were caught outside at the start of the experiment ranged more throughout its duration (caught outside: 72%, caught inside 51%, SEM: 4, $\chi_1^2 = 10.0$, P = 0.002), and individual range use was correlated between monitoring cycles (for adjacent monitoring cycles: $r_s^2 = 0.5-0.7$, P < 0.0001). This emphasizes the importance of studying range use on an individual level. In conclusion, our light-based monitoring system can assess individual range use accurately (although accuracy was affected by house characteristics to some extent) and was used to show that both cover availability and individual characteristics affected range use.