The identification of social interactions is of fundamental importance for animal behavioral studies, addressing numerous problems like investigating the influence of social hierarchical structures or the drivers of agonistic behavioral disorders. However, the majority of previous studies often rely on manual determination of the number and types of social encounters by direct observation which requires a large amount of personnel and economical efforts. To overcome this limitation and increase research efficiency and, thus, contribute to animal welfare in the long term, we propose in this study a framework for the automated identification of social contacts. In this framework, we apply a convolutional neural network (CNN) to detect the location and orientation of pigs within a video and track their movement trajectories over a period of time using a Kalman filter (KF) algorithm. Based on the tracking information, we automatically identify social contacts in the form of head–head and head–tail contacts. Moreover, by using the individual animal IDs, we construct a network of social contacts as the final output. We evaluated the performance of our framework based on two distinct test sets for pig detection and tracking. Consequently, we achieved a Sensitivity, Precision, and F1-score of 94.2%, 95.4%, and 95.1%, respectively, and a $MOTA$ score of 94.4%. The findings of this study demonstrate the effectiveness of our keypoint-based tracking-by-detection strategy and can be applied to enhance animal monitoring systems.

Behavioural research of pigs can be greatly simplified if automatic recognition systems are used. Systems based on computer vision in particular have the advantage that they allow an evaluation without affecting the normal behaviour of the animals. In recent years, methods based on deep learning have been introduced and have shown excellent results. Object and keypoint detector have frequently been used to detect individual animals. Despite promising results, bounding boxes and sparse keypoints do not trace the contours of the animals, resulting in a lot of information being lost. Therefore, this paper follows the relatively new approach of panoptic segmentation and aims at the pixel accurate segmentation of individual pigs. A framework consisting of a neural network for semantic segmentation as well as different network heads and postprocessing methods will be discussed. The method was tested on a data set of 1000 hand-labeled images created specifically for this experiment and achieves detection rates of around 95% (F1 score) despite disturbances such as occlusions and dirty lenses.

This study aimed to investigate how farrowing and rearing systems affect skin lesions, serum cortisol, and aggressive behavior as indicators for weaning stress of piglets. Between May 2016 and March 2018, in total 3144 weaning piglets from three different farrowing systems were examined: farrowing crates (FC), single-housing free-farrowing pens (FF), and group-housing of lactating sows and litters (GH). After weaning and regrouping, piglets were relocated to conventional rearing pens (conv; 5.7 m²) or to wean-to-finish pens (w-f; 12.4 m²). Skin lesions were scored 24 h after weaning. Blood samples were taken one week before and 24 h after weaning to analyze the individual difference in serum cortisol. Behavior was observed for 24 h after relocation. Animals raised in FC and FF had significantly more skin lesions than that of GH animals. Piglets born in GH showed lower cortisol differences and fought less and for shorter periods compared to FC and FF piglets. Piglets weaned to w-f pens showed greater cortisol changes and fought significantly longer than piglets in conv pens. Group housing during the suckling period reduced weaning stress for piglets in terms of skin lesions, serum cortisol, and aggressive behavior. Greater space allowance (w-f vs. conv) was not beneficial with regard to the investigated parameters.

This study aimed to investigate the effects of farrowing and rearing systems on tail lesions and losses of docked and undocked pigs. Pigs from three farrowing systems: Conventional farrowing crate (FC), free farrowing (FF) and group housing of lactating sows (GH) were randomly allocated to different rearing systems: A conventional system (CONV), where the pigs were regrouped and transferred to conventional finishing pens at ten weeks of age or a wean-to-finish (W-F) system, where the pigs remained in their pens until slaughter with higher space allowance during rearing. Weekly, tail lesions and losses were assessed individually. The incidence of tail lesions was higher in undocked CONV pigs compared to undocked W-F pigs (maximum: CONV 58.01%, W-F 41.16%). The rearing system had a significant effect on tail losses at the end of finishing (CONV 67.63%, W-F 38.2%). The significant effect of the rearing system might be explained by higher space allowance during rearing and reduced regrouping stress for W-F pigs. In conclusion, farrowing systems showed no effects, but the W-F rearing system reduces the frequency of tail lesions and losses; the curves of tail lesions increased slower and stayed on a lower level, which resulted in lower losses as well.

The aim of the present study was to classify and characterise pigs with tail lesions using a combined parameter based on the frequency and duration of tail lesions and to find out whether biologically relevant groups could be separated by cluster analysis. Pigs (n = 677, 50% docked, 50% undocked) from three farrowing systems, as follows: (1) Conventional farrowing crate (FC), (2) free farrowing (FF), and (3) a group housing lactating sows (GH), were divided into two rearing systems as follows: (1) A conventional system (CONV) and (2) a wean-to-finish (W-F) system. Within 18 assessment weeks, starting after weaning, animal tail lesions were recorded individually. The animals were characterised into five lesion groups, as follows: (I) No lesions to (V) many long lasting lesions. The separability of the predefined lesion groups was checked by an animal individual lesion parameter. By using a k-means cluster analysis, it was shown that the docking status was the mainly affected parameter on the tail lesions. The separation of the groups only succeeded for the most distinct groups, I and V. The high impact of the docking status and the reduction of tail lesions by more space allowance was shown. More characterising information for the individual pigs would improve the separability of the lesion groups.

The behavioural research of pigs can be greatly simplified if automatic recognition systems are used. Especially systems based on computer vision have the advantage that they allow an evaluation without affecting the normal behaviour of the animals. In recent years, methods based on deep learning have been introduced and have shown pleasingly good results. Especially object and keypoint detectors have been used to detect the individual animals. Despite good results, bounding boxes and sparse keypoints do not trace the contours of the animals, resulting in a lot of information being lost. Therefore this work follows the relatively new definition of a panoptic segmentation and aims at the pixel accurate segmentation of the individual pigs. For this a framework of a neural network for semantic segmentation, different network heads and postprocessing methods is presented. With the resulting instance segmentation masks further information like the size or weight of the animals could be estimated. The method is tested on a specially created data set with 1000 hand-labeled images and achieves detection rates of around 95% (F1 Score) despite disturbances such as occlusions and dirty lenses.