ORGAN LOCALIZATION AND DETECTION IN SOW’S USING MACHINE LEARNING AND DEEP LEARNING IN COMPUTER VISION

Aug 4, 2026

Upload by Aurora

Automated farming and animal husbandry increasingly rely on computer vision to reduce costly expert involvement, yet accurate assistance for sow insemination outcomes remains challenging. This thesis investigates machine learning and deep learning methods for object detection and segmentation using thermal imagery, with a focus on detecting and localizing the sow’s vulva. The work evaluates traditional semi-supervised approaches on infrared images, proposes a new feature extractor by combining two extractors, and applies deep learning pipelines to support automated downstream estrus-cycle and ovulation estimation.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages69

Reading Time2 h 54 min

AccessFree to download

Acknowledgements

Abbreviations

Thesis Statement and Contribution Summary

List of Figures

List of Tables

Abstract

1 Introduction

2 Object Detection Using Machine Learning

2.1 Background

2.2 Method Overview

3 Object Detection Implementation Using Deep Learning

3.1 Background

3.2 Method overview

4 Image Segmentation Using U-Net

4.1 Background

4.2 Introduction

5 Thesis Results

5.1 Results of machine learning model

University of Memphis University of Memphis Digital Commons Electronic Theses and Dissertations 5-2-2023 ORGAN LOCALIZATION AND DETECTION IN SOW’S USING MACHINE LEARNING AND DEEP LEARNING IN COMPUTER VISION Iyad Almadani Follow this and additional works at: [https://digitalcommons.memphis.edu/etd](https://digitalcommons.memphis.edu/etd) Recommended Citation Almadani, Iyad, "ORGAN LOCALIZATION AND DETECTION IN SOW’S USING MACHINE LEARNING AND DEEP LEARNING IN COMPUTER VISION" (2023) . Electronic Theses and Dissertations. 3104. [https://digitalcommons.memphis.edu/etd/3104](https://digitalcommons.memphis.edu/etd/3104) This Thesis is brought to you for free and open access by University of Memphis Digital Commons. It has been accepted for inclusion in Electronic Theses and Dissertations by an authorized administrator of University of Memphis Digital Commons. For more information, please contact [khggerty@memphis.edu](khggerty@memphis.edu). ORGAN LOCALIZATION AND DETECTION IN SOW’S USING MACHINE LEARNING AND DEEP LEARNING IN COMPUTER VISION by Iyad Mohd Eid H Almadani A Thesis Submitted in Partial Fulfillment of the Requirements for the Degree of Master of Science Major: Electrical And Computer Engineering The University of Memphis May 2023 ©IYAD ALMADANI, 2023 i Acknowledgements First and foremost, All praises and gratitude to Allah who gives me the strength to accomplish this degree. This thesis is dedicated to my sacrificer mother, my supportive brother Murad as well as my educator and virtuous father, MOHAMMAD EID, who has passed away(may Allah elevate his rank in paradise) . Additionally, I want to express my most profound appreciation to my friend Mohammed Abu Hussein, and I am also indebted to my supervisor Dr. Aaron L Robinson for always having my back. Abbreviations | Abbreviations | | | --- | --- | | Abbreviation | Meaning | | SVM
HOG
PCA
CNN
YOLO
HSV
IOU
LWIR
ROI | support vector machine Histogram of Oriented Gradients Principle Component Analysis
Convolutional Neural Network
You Only Look Once
Hue, Saturation, and value
Intersection Over Union
Long-Wave Infrared
Regions Of Interest | Thesis Statement and Contribution Summary Farming and animal husbandry has benefited from recent developments in automated harvesting, scientific studies in plant genomes, and recent reproductive cycle research. However, even with these recent advances, the industry is still heavily dependent upon human specialists with many years of experience to limit the costs associated with insemination and to maximize the probability of pregnancy associated with insemination. To address this issue, this thesis poses the question of how can the dependence on human experts be decreased through the application of machine learning techniques? To begin to answer this question, this treatment focuses on object detection and segmentation using thermal images. Specifically, this research focuses on the problem of detecting the sow’s vulva. This thesis also poses the question of whether current research results can contribute to an automated process of determining the estrus cycle and ovulation in sows? This thesis assembles and presents research and results that begin to address these questions. The contributions of this thesis may be summarized as follows: 1. Development of machine learning techniques to reduce the farming industry dependence on human experts. 2. Development of segmentation results for vulva detection and localization. 3. Derivation of a new feature extractor based on concatenating two feature extractors. 4. Evaluation of traditional semi-supervised ML techniques on infrared images Table of Contents Acknowledgements .......................................... ii Abbreviations ............................................. iii Thesis Statement and Contribution Summary ............................ iv List of Figures ........

FAQs

What problem does the thesis address in automated sow management?

It targets reducing dependence on human specialists by automating organ localization and detection for insemination-related decision support, specifically focusing on vulva detection from thermal images.

Which imaging modality is used for detection and segmentation?

Thermal imagery is used, including infrared images, to detect and segment the sow’s vulva and related regions.

What methods and models are evaluated in the thesis?

The research covers traditional machine learning for object detection, deep learning-based detection workflows, and U-Net for image segmentation, alongside evaluation of semi-supervised ML techniques on infrared data.

Aug 4, 2026

Upload by Aurora