Automated Visual Sensemaking of Swine Respiration Rates

Thumbnail Image
Date
2022-12
Authors
Handa, Divya
Major Professor
Advisor
Peschel, Joshua
Ramirez, Brett
Johnson, Anna
Koziel, Jacek
Hansen, Stephanie
Committee Member
Journal Title
Journal ISSN
Volume Title
Publisher
Authors
Research Projects
Organizational Units
Journal Issue
Is Version Of
Versions
Series
Department
Agricultural and Biosystems Engineering
Abstract
Respiration rate (RR) of livestock is a vital parameter that can give an insight into the livestock’s health and well-being. Livestock respiration is commonly assessed visually by observing abdomen fluctuation; however, the traditional methods are time consuming, subjective, being therefore impractical for large-scale operations and must rely on automation. Automatic monitoring of livestock provides an opportunity to continuously monitor them without causing any hindrance to the farm activities. Continuous monitoring of livestock enables the producers to observe subtle behavioral changes which could be easily missed by method of direct human observation. Contact and non-contact technologies are used to automatically monitor respiration rate; contact technologies (e.g., accelerometers, pressure sensors, and thermistors) utilize sensors that are physically mounted on livestock while non-contact technologies (e.g., machine vision, thermography, and sound analysis) enable a non-invasive method of monitoring respiration. Among all the new technologies, machine vision technology is becoming widely popular. The machine vision technology relies on the use of cameras to acquire video or image data. The video/ image data is used to train the model. The models could be trained to identify object, perform segmentation, detect abnormal behaviors etc. Abnormal respiration rate could be an indicator of potential respiratory disease or heat stress. In this study, machine vision technology has been adopted to monitor respiration rate of the pigs in a commercial production system. The overall aim is to develop and apply machine vision algorithms for use in real world environments. The methodology is divided into three stages. First, an object detection model Mask R-CNN is employed to detect and segment pigs from the background. Second, Phase Based Video Magnification (PBVM) algorithm is used to amplify weak respiratory movements of lying resting pigs. Third, peak to peak estimation approach is adopted to estimate the respiration rate from the amplified signal. The respiration rate obtained using peak to peak estimation was then compared with manual respiration rate estimates to validate the performance of the algorithm. The proposed methodology provides an accurate and reliable alternative to human observation. Implementing machine vision for respiration estimation reduces the labor costs, saves time, and provides an opportunity for timely intervention. The beneficiaries of this study include livestock researchers and practitioners. The practitioners can improve the health and welfare of their livestock and reduce the economic losses. The livestock researchers could use the results of this study to explore the application of the methodology on other livestock. The results of this study also provide recommendations for future studies.
Comments
Description
Keywords
Citation
DOI
Source
Subject Categories
Copyright