Automatic Behavior and Posture Detection of Sows in Loose Farrowing Pens Based on 2D-Video Images

Abstract

The monitoring of farm animals and the automatic recognition of deviant behavior have recently become increasingly important in farm animal science research and in practical agriculture. The aim of this study was to develop an approach to automatically predict behavior and posture of sows by using a 2D image-based deep neural network (DNN) for the detection and localization of relevant sow and pen features, followed by a hierarchical conditional statement based on human expert knowledge for behavior/posture classification. The automatic detection of sow body parts and pen equipment was trained using an object detection algorithm (YOLO V3). The algorithm achieved an Average Precision (AP) of 0.97 (straw rack), 0.97 (head), 0.95 (feeding trough), 0.86 (jute bag), 0.78 (tail), 0.75 (legs) and 0.66 (teats). The conditional statement, which classifies and automatically generates a posture or behavior of the sow under consideration of context, temporal and geometric values of the detected features, classified 59.6% of the postures (lying lateral, lying ventral, standing, sitting) and behaviors (interaction with pen equipment) correctly. In conclusion, the results indicate the potential of DNN toward automatic behavior classification from 2D videos as potential basis for an automatic farrowing monitoring system.

Implementation of Computer-Vision-Based Farrowing Prediction in Pens with Temporary Sow Confinement

“…In the study of Küster et al [ 25 ], YOLOv3 was used to detect different parts of sows’ bodies in farrowing pens, i.e., heads, tails, legs, and udder, but not the whole sows’ bodies, as in our study. They detected heads with 97 AP 50 , tails with 78 AP 50 , legs with 75 AP 50 and udder with 66 AP 50 .…”

“…In our study, the speed of YOLOX was from 21 fps with YOLOX-extra large to 42 fps with YOLOX-nano. A comparison to YOLOv3 in the research of Küster et al [ 25 ] and van der Zande et al [ 26 ] was not possible as the speed of inference of YOLOv3 was not reported by the authors.…”

“…In both studies, the first of Küster et al [ 25 ] and the second of van der Zande et al [ 26 ], the validation of object detection models was performed on the same pens as the training of the models. This revealed the performance of the trained models in these pens.…”

Pose Estimation of Sow and Piglets During Free Farrowing Using Deep Learning

“…Also a multi-pig trajectory tracking model based on DeepLabCut was described, but only in younger growing pigs [42]. Images were collected from either crated sows [38,43,45] or loose-housed sows [8,44] either before farrowing [46], during farrowing [38] or during lactation [8,[43][44][45]. In these studies, either a 2D RGB-camera [38,44,46] or 3D Kinetic camera [8,43,45] was used.…”

Detecting Animal Contacts—A Deep Learning-Based Pig Detection and Tracking Approach for the Quantification of Social Contacts

“…Here, the most frequently used detection approaches aim to localize an object of interest by computing a bounding box around the object [19][20][21]. Although these approaches work successfully for various problem settings, due to the overlapping of the predicted bounding boxes, their applicability is limited for the analysis of videos with high utilization rates and several pigs in a close environment [22,23]. Furthermore, the standard bounding box approach only provides the positional information without taking the orientation of the animal into account which is a key information in order to reliably differentiate distinctive contact types like head-head interactions.…”