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

Martin Wutke,Imke Traulsen,Pronaya Prosun Das,Armin Otto Schmitt,Mehmet Gültas,Maria Gentz,Friederike K Warns,Felix Heinrich,Anita Lange

doi:10.3390/s21227512

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 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.

Highlights

Today, it is well known that domestic pigs are highly social animals, maintaining hierarchical structures and socially organized groups
In order to assess the overall performance of our framework, we evaluated both the convolutional neural network (CNN) detection stage as well as the multi-target pig tracking stage separately
We focused on the tracking ability of the implemented Kalman filter (KF) and used 70 randomly selected video sequences as the tracking evaluation data

Summary

Introduction

It is well known that domestic pigs are highly social animals, maintaining hierarchical structures and socially organized groups. The established social orders are frequently disrupted due to mixing groups as they are transferred between different housing and production stages [1,2]. In order to enhance animal welfare and health in future husbandry systems, the analysis of animal interactions as well as their monitoring and prediction is of high importance in research and commercial farming. Within the area of precision livestock farming, the tasks of multiple object detection and motion tracking have been studied intensively in recent years, in order to remotely monitor several animals and to capture the animals activity [11,12,13,14,15]. While multiple object detection refers to the task of locating several objects belonging to a category of interest within an image [16], multiple object tracking can be described as tracing the movement of objects throughout a consecutive number of video frames and consistently assigning individual object IDs [17]

Objectives

Methods

Results

Conclusion