A Kinect-Based Segmentation of Touching-Pigs for Real-Time Monitoring.

Miso Ju,Daihee Park,Younchang Choi,Jihyun Seo,Jaewon Sa,Sungju Lee,Yongwha Chung

doi:10.3390/s18061746

Abstract

Segmenting touching-pigs in real-time is an important issue for surveillance cameras intended for the 24-h tracking of individual pigs. However, methods to do so have not yet been reported. We particularly focus on the segmentation of touching-pigs in a crowded pig room with low-contrast images obtained using a Kinect depth sensor. We reduce the execution time by combining object detection techniques based on a convolutional neural network (CNN) with image processing techniques instead of applying time-consuming operations, such as optimization-based segmentation. We first apply the fastest CNN-based object detection technique (i.e., You Only Look Once, YOLO) to solve the separation problem for touching-pigs. If the quality of the YOLO output is not satisfied, then we try to find the possible boundary line between the touching-pigs by analyzing the shape. Our experimental results show that this method is effective to separate touching-pigs in terms of both accuracy (i.e., 91.96%) and execution time (i.e., real-time execution), even with low-contrast images obtained using a Kinect depth sensor.

Highlights

When caring for group-housed livestock, the early detection and management of problems related to health and welfare is important
A guideline could be generated in the natural hole and connected to the concave points (CPs) or the nearest points, which would be located on the outline of the
With a straightforward method, touching-pigs cannot be separated accurately in real-time due to the relatively low-contrast images obtained from a Kinect depth sensor

Summary

Introduction

When caring for group-housed livestock, the early detection and management of problems related to health and welfare is important. The care of individual animals is necessary to minimize the potential damage caused by infectious disease or other health and welfare problems. It is almost impossible for a small number of farm workers to care for individual animals on a large livestock farm. The present study contributes to our final goal of 24-h automatic pig behavior analysis by focusing on identifying individual pigs based on pig segmentation. The key problem when segmenting and tracking weaning pigs continuously during the automatic analysis of pig behavior is to separate touching-pigs in a crowded environment

Objectives

Methods

Results

Discussion

Conclusion