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

Abstract

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.

Industry 4.0 and Precision Livestock Farming (PLF): An up to Date Overview across Animal Productions

“…This approach was applied as a convolutional neural network to detect the orientation of pigs and to follow their movements using a Kalman filter algorithm. This system was able to construct a network of social connections and, as a result, evaluate the pigs' behaviour with a sensitivity of 94.2% and a precision of 95.4% [171]. Other studies recognised aggressive episodes of pigs by processing of picture and video material from pens [172,173].…”

An efficient anchor‐free method for pig detection

“…Wutke et al. [12] also detect the lines connecting the tail and shoulder, left shoulder and ear, and right shoulder and ear of each animal. However, the labelling process requires additional time, because each part (i.e.…”

“…Psota et al [6] suggest individually detecting the parts of the animal and then grouping them together using an association algorithm. Wutke et al [12] also detect the lines connecting the tail and shoulder, left shoulder and ear, and right shoulder and ear of each animal.…”

“…Despite the good performance of their datasets, these detection algorithms have a poor generalization capability. The use of convolutional neural networks (CNNs) for computer vision tasks for pig monitoring has become increasingly common with recent advances in technology and deep learning [2][3][4][5][6][7][8][9][10][11][12]. While this technology has shown satisfactory results for the detection of other animals and humans [13], the various constraints mentioned above prevent the development of a robust framework for pig detection under real-world conditions.…”

“…Furthermore, a region with an occluded pig might not contain sufficient features to be classified as positive. A few studies suggest a bottom-up approach to detect pigs in an image [6,7,12]. They detect the individual animal parts and use an association algorithm to group them together.…”