Keypoint Detection for Injury Identification during Turkey Husbandry Using Neural Networks.

Injurious pecking against conspecifics is a serious problem in turkey husbandry. Bloody injuries act as a trigger mechanism to induce further pecking, and timely detection and intervention can prevent massive animal welfare impairments and costly losses. Thus, the overarching aim is to develop a camera-based system to monitor the flock and detect injuries using neural networks. In a preliminary study, images of turkeys were annotated by labelling potential injuries. These were used to train a network for injury detection. Here, we applied a keypoint detection model to provide more information on animal position and indicate injury location. Therefore, seven turkey keypoints were defined, and 244 images (showing 7660 birds) were manually annotated. Two state-of-the-art approaches for pose estimation were adjusted, and their results were compared. Subsequently, a better keypoint detection model (HRNet-W48) was combined with the segmentation model for injury detection. For example, individual injuries were classified using "near tail" or "near head" labels. Summarizing, the keypoint detection showed good results and could clearly differentiate between individual animals even in crowded situations.

Footpad Monitoring: Reliability of an Automated System to Assess Footpad Dermatitis in Turkeys (Meleagris gallopavo) During Slaughter.

Footpad dermatitis (FPD) is an indicator of animal welfare in turkeys, giving evidence of the animals' physical integrity and providing information on husbandry management. Automated systems for assessing FPD at slaughter can present a useful tool for objective data collection. However, using automated systems requires that they reliably assess the incidence. In this study, the feet of turkeys were scored for FPD by both an automated camera system and a human observer, using a five-scale score. The observer reliability between both was calculated (Krippendorff's alpha). The results were not acceptable, with an agreement coefficient of 0.44 in the initial situation. Therefore, pictures of 3,000 feet scored by the automated system were evaluated systematically to detect deficiencies. The reference area (metatarsal footpad) was not detected correctly in 55.0% of the feet, and false detections of the alteration on the footpad (FPD) were found in 32.9% of the feet. In 41.3% of the feet, the foot was not presented straight to the camera. According to these results, the algorithm of the automated system was modified, aiming to improve color detection and the distinction of the metatarsal footpad from the background. Pictures of the feet, now scored by the modified algorithm, were evaluated again. Observer reliability could be improved (Krippendorff's alpha = 0.61). However, detection of the metatarsal footpad (50.9% incorrect detections) and alterations (27.0% incorrect detections) remained a problem. We found that the performance of the camera system was affected by the angle at which the foot was presented to the camera (skew/straight; p < 0.05). Furthermore, the laterality of the foot (left/right) was found to have a significant effect (p < 0.001). We propose that the latter depends on the slaughter process. This study also highlights a high variability in observer reliability of human observers. Depending on the respective target parameter, the reliability coefficient (Krippendorff's alpha) ranged from 0.21 to 0.82. This stresses the importance of finding an objective alternative. Therefore, it was concluded that the automated detection system could be appropriate to reliably assess FPD at the slaughterhouse. However, there is still room to improve the existing method, especially when using FPD as a welfare indicator.

Structuring Broiler Barns: How a Perforated Flooring System Affects Animal Behavior.

Broiler chickens in Europe are usually raised in a barren environment. Elevated perforated platforms address this problem and can positively influence animal health and welfare. To evaluate the effect of an elevated perforated floor on the behavior of broiler chickens, one of two barns was equipped with a perforated flooring system under the food and water supply. The second barn was used as a control. In total, three fattening periods were observed, with 500 broiler chickens (Ross 308 breed) kept in each barn. To compare the behavior of the birds in these groups, cameras were installed in the two barns. The videos were analyzed by counting the number of birds and observing focal animals while recording their behavior. More animals were observed on the perforated floor than in the littered control area (p < 0.001), but focal animals spent less time on the perforated floor compared to the observed littered area in the control barn (p < 0.05). There were no differences in the length of the recorded behaviors between the treatments. These findings suggest that, in general, the elevated perforated floor is attractive for the animals. However, it does not promote one of the recorded behavior patterns. Our results show that an elevated perforated floor could be an option for structuring broiler barns.

Learn to Train: Improving Training Data for a Neural Network to Detect Pecking Injuries in Turkeys.

This study aimed to develop a camera-based system using artificial intelligence for automated detection of pecking injuries in turkeys. Videos were recorded and split into individual images for further processing. Using specifically developed software, the injuries visible on these images were marked by humans, and a neural network was trained with these annotations. Due to unacceptable agreement between the annotations of humans and the network, several work steps were initiated to improve the training data. First, a costly work step was used to create high-quality annotations (HQA) for which multiple observers evaluated already annotated injuries. Therefore, each labeled detection had to be validated by three observers before it was saved as "finished", and for each image, all detections had to be verified three times. Then, a network was trained with these HQA to assist observers in annotating more data. Finally, the benefit of the work step generating HQA was tested, and it was shown that the value of the agreement between the annotations of humans and the network could be doubled. Although the system is not yet capable of ensuring adequate detection of pecking injuries, the study demonstrated the importance of such validation steps in order to obtain good training data.

Effects of a slatted floor on bacteria and physical parameters in litter in broiler houses.

In this study, a new housing system for broiler was tested. This system consisted of a slatted floor area and a littered area with the aim of improving litter quality. Two experimental broiler houses were provided. In house 1, a slatted floor was installed below the drinker and feedlines. Littered areas flanked the slatted floor. Broiler house 2 reflected conditions in commercial systems, consisting of a full littered area. Litter samples were taken at day 11 and at day 32 of the fattening period. Manure samples were taken at day 32. The total bacteria count (TBC), coliforms, Escherichia coli (E. coli) and ESBL-producing bacteria were determined. Furthermore, physical parameters (dry matter, water activity, pH) of litter and manure were measured. For statistical analyzes, a generalized linear mixed model (GLIMMIX procedure) was calculated. The floor did not show any significant effect on the bacteria content of the litter. Regarding TBC in litter, the floor showed a tendency for an effect (F = 5.42, p<0.1) with lower contents in house 1. Regarding the manure under the slatted floor, a tendency for a difference between house 1 and house 2 was found for the content of E. coli (F = 5.55, p<0.1) with higher contents in house 1. The floor did not show any significant effect on the physical parameters of litter and manure. The results of this experimental study showed no positive effects on the selected litter parameters, but further studies, especially on-farm experiments are necessary to confirm these results.

Walking on Tiptoes: Digital Pads Deserve Increased Attention When Scoring Footpad Dermatitis as an Animal Welfare Indicator in Turkeys.

Animal welfare is one of the most challenging issues in modern farm animal husbandry. Animal welfare indicators can be used to monitor welfare on farms or at slaughterhouses, with footpad dermatitis (FPD) being one of the most important indicators used in turkeys. Up to now, the severity of FPD has been measured by evaluating the size of altered lesions on the metatarsal pad of birds. However, such lesions are not only found on the metatarsal pads, but alterations can also occur on the digital pads of the animals, the latter is not included in the European standard scoring systems for turkeys so far. The aim of the present study was to give a detailed outline of alterations on the digital pads of turkeys and associate their occurrence to a standardly used five-point scoring system, which is based on alterations of the metatarsal pad only. Therefore, pictures of 500 feet of turkeys from 16 flocks at the end of the fattening phase were taken, using an automatic camera system. Based on these pictures, alterations on the digits were scored according to different parameters (lesions, swellings, and number of affected digits). Furthermore, detailed measurements were conducted using an imaging software. Results were compared with a standardly used five-point scoring system (standard FPD scoring system), based on the metatarsal pad as reference. Results provide no equivalence in occurrence and severity of alterations on the metatarsal pads compared to those found on the digits. Pathologic alterations on the digits were already present at standard FPD scoring level 0; no differentiation became obvious between the higher scoring levels 2-4. Strong correlations were found when comparing percentage of alterations of the standard FPD scoring system to those of a system including alterations on the digits and the metatarsal pad, using the total foot as a reference (rp = 0.9, p < 0.001). This was the first study conducting a detailed analysis of alterations on the digits of turkeys. In conclusion, results of this study show that including the evaluation of alterations on digits could refine the present FPD scoring system, especially when using FPD as an animal welfare indicator.

Effects of Enriched Charcoal as Permanent 0.2% Feed-Additive in Standard and Low-Protein Diets of Male Fattening Turkeys: An On-Farm Study.

Wet litter is the most important cause of footpad dermatitis in poultry, this in turn being a highly relevant animal-related welfare indicator. This field study was subdivided into two experiments. In Experiment 1, the standard diet was supplemented by 0.2% enriched charcoal, being a non-specific absorber and therefore might be promising in reducing faecal moisture. In Experiment 2, the experimental group received a reduced crude protein diet during weeks 6-13, combined with a 0.2% enriched charcoal supplementation. The trials were each conducted with two batches on three farms under on-farm conditions. The animals were observed at 6, 10, 14 and 18 weeks of age to collect data on body weight and different health parameters. The mortality and litter samples were analysed after slaughtering. In Experiment 1, performance and health were not affected despite higher dry matter content of the litter. In Experiment 2, the weight of birds receiving the protein-reduced diet was decreased significantly throughout the experiment. However, the slaughter weight did not differ. The mortality was reduced by 0.5% in the experimental group. Therefore, it was concluded that 0.2% of enriched charcoal is not a valuable feed-additive regarding animal health, while temporary protein reduction might have positive effects.

Evaluation of a gait scoring system for cattle by using cluster analysis and Krippendorff's α reliability.

The aim of the presented study was to validate a three-point locomotion score (LS) classifying lameness in dairy cows. Therefore, locomotion of 144 cows was scored and data on claw lesions were collected during hoof trimming. Based on latter data a cluster analysis was performed to objectively classify cows into three groups (Cluster 1-3). Finally, the congruence between scoring system and clustering was tested using Krippendorff's α reliability. In total, 63 cows (43.7 per cent) were classified as non-lame (LS1), 38 (26.4 per cent) were rated as LS2 with an uneven gait and 43 (29.9 per cent) cows were ranked as clearly lame (LS3). In comparison, hoof-trimming data revealed 64 cows (44.4 per cent) to show no diagnosis, 37 (25.7 per cent) one diagnosis, 33 animals (22.9 per cent) two diagnoses and 10 (7.0 per cent) more than two. Comparing the respective categorisation received by either the cluster analysis or LS in between groups, a high correspondence (79.4 per cent and 83.7 per cent) could be found for LS1 and cluster 1 as well as for LS3 and cluster 3. Only LS2 had partial agreement (21.1 per cent) to cluster 2. However, Krippendorff's α was 0.75 (95 per cent CI 0.68 to 0.81), indicating a good degree of reliability. Therefore, the results of this study suggested that the presented LS is suitable for classifying the cows' state of lameness representing their claw diseases.

First approach validating a scoring system for foot-pad dermatitis in broiler chickens developed for application in practice.

Measuring the severity of foot-pad dermatitis is an accepted tool monitoring the quality of animal husbandry and welfare. Up to now, a variety of scoring schemes have been used, most of them based on visual evaluation. However, a standardisation and validation of scoring systems is beneficial, not only to compare different studies, but to provide objective indicators in poultry welfare. In this study, we validated one visual scoring system, widely used in Northern Germany, by using additional information of histological measurements. Therefore, feet of broiler chickens (ROSS 308) from one flock were visually scored at the slaughter plant (4-point score). Ten feet per score level (n = 40) were sampled and analysed macroscopically and microscopically. Data were analysed using cluster analysis, providing a classification based on these histopathological findings. Validity of the visual scoring system was analysed by (1) testing the interobserver reliability between different observers and (2) by comparing both, visual and cluster classification types using the McNemar's test. In a last step Kendall tau correlations were calculated in order to find suitable parameters to judge the severity in a visual score more reliably. Results could show that most agreement was found for the score levels 1 and 2, whereas results for score levels 3 and 4 were more divergent. These results were found in both, interobserver reliability and comparison of classification types (visual vs. cluster). Results revealed interaction effects of classification type and scoring level for the width of ulcers (p = 0.0044) and the size of the lesion (p = 0.0081). In the cluster classification, higher values in both, width of ulcer and size of lesion could be found in score level 3. Furthermore, a positive correlation of the size of lesion with the depth of the ulcer was found (0.73). In conclusion, we found that histological findings coincided well with the less severe visual scores (1; 2), whereas the differentiation between the severe scores (3; 4) seemed to be less valid. For practical purposes we therefore recommend keeping visual scoring systems simple. Furthermore, as the correlation coefficient between both was quite high, the size of the lesion might serve as an indirect indicator of the depth.

Serotonin depletion induces pessimistic-like behavior in a cognitive bias paradigm in pigs.

Cognitive and affective processes are highly interrelated. This has implications for neuropsychiatric disorders such as major depressive disorder in humans but also for the welfare of non-human animals. The brain serotonergic system might play a key role in mediating the relationship between cognitive functions and affective regulation. The aim of our study was to examine the influence of serotonin depletion on the affective state and cognitive processing in pigs, an important farm animal species but also a potential model species for biomedical research in humans. For this purpose, we modified a serotonin depletion model using para-chlorophenylalanine (pCPA) to decrease serotonin levels in brain areas involved in cognitive and affective processing (part 1). The consequences of serotonin depletion were then measured in two behavioral tests (part 2): the spatial judgement task (SJT), providing information about the effects of the affective state on cognitive processing, and the open field/novel object (OFNO) test, which measures behavioral reactions to novelty that are assumed to reflect affective state. In part 1, 40 pigs were treated with either pCPA or saline for six consecutive days. Serotonin levels were assessed in seven different brain regions 4, 5, 6, 11 and 13days after the first injection. Serotonin was significantly depleted in all analyzed brain regions up to 13days after the first application. In part 2, the pCPA model was applied to 48 animals in behavioral testing. Behavioral tests, the OFNO test and the SJT, were conducted both before and after pCPA/saline injections. While results from the OFNO tests were inconclusive, an effect of treatment as well as an effect of the phase (before and after treatment) was observed in the SJT. Animals treated with pCPA showed more pessimistic-like behavior, suggesting a more negative affective state due to serotonin depletion. Thus, our results confirm that the serotonergic system is a key player in cognitive-emotional processing. Hence, the serotonin depletion model and the spatial judgement task can increase our understanding of the basic mechanisms underlying both human neuropsychiatric disorders and animal welfare.

Assessment of stress in laboratory beagle dogs constrained by a Pavlov sling.

The 3Rs - Replacement, Reduction and Refinement - have become increasingly important in designing animal experiments. The Pavlov sling is thought to be a non-invasive method to restrain dogs for examinations. The aim of our study was to investigate whether laboratory Beagle dogs that had been trained to tolerate restraint by a Pavlov sling are stressed by this procedure and, furthermore, to analyze their behavior during this period. Five male and five female Beagle dogs were used, each three years of age. Animals were restrained in the Pavlov sling for 30 min on six days with an interval of at least two days. The following behaviors were recorded every minute for each session: postures of body, head, and ears, as well as state of eyes, tail, legs, and mouth. Additionally, the animals were observed for the occurrence of particular stress signs, including body shaking, sweating of the paws, increased saliva production, piloerection, blinking of eyes, snout licking, yawning, and panting. As an indicator for stress, salivary cortisol levels were measured before, during, and after each session. Our results show that for most behavioral parameters, e.g., body, leg, head, tail, and ear posture, the frequency of changes between different behavior patterns, as well as cortisol concentration, were not influenced by restraint in the Pavlov sling. Therefore, the Pavlov sling does not seem to be perceived as a stressful situation by the Beagle dogs. Our study demonstrates that under certain conditions the use of the Pavlov sling in trained dogs can substitute for more ordinary methods of immobilization, e.g., the use of narcotics.