Go to main content
Formats
Format
BibTeX
MARCXML
TextMARC
MARC
DataCite
DublinCore
EndNote
NLM
RefWorks
RIS

Files

Abstract

With rising consumer concerns about animal welfare, the United States (USA) egg industry is shifting towards cage-free farming practices. This shift introduces challenges in poultry management, sustainable egg production, and automation in poultry farming. In response, this study investigates updated computer vision techniques, thermal cameras, and robotics to monitor poultry floor distribution, predict bird body weight, manage floor eggs, and detect behaviors and welfare conditions. The objective of this dissertation was to evaluate the performance of traditional convolutional neural network (CNN) models (e.g., YOLO series, EfficientNetV2, SegFormer, and SETR) and large vision models (LVM) (e.g., Segment Anything Model and Track Anything Model) in assessing key production and welfare indicators of cage-free layers. Furthermore, the research explored advanced robotic systems for detecting floor eggs and dead chickens. For this study, 800 hens were raised in four cage-free research rooms under different experimental designs based on specific research objectives. The results demonstrated that CNN models can effectively track chickens' spatial distribution (90.0% precision), detect floor eggs (94.8% accuracy), and classify six behaviors (i.e., feeding, drinking, walking, perching, dust bathing, and nesting) with 95.3% accuracy. LVMs, combined with thermal cameras, predicted chicken body weight (R² = 0.90) and tracked individual hens (RMSE = 0.02 m/s). Moreover, integrating CNN models with intelligent bionic quadruped robots allowed for the detection of floor eggs in dimly lit areas, such as beneath feeders and in corner spaces, as well as the identification of dead chickens within the flock. In conclusion, this dissertation highlights the cutting-edge techniques of precision farming technologies in advancing automated poultry management in cage-free systems. By integrating CNN, LVM, and robotic technologies, this research offers an interdisciplinary approach to addressing the challenges of modern cage-free farming, advancing the poultry industry with more ethical, efficient, automated, and sustainable production practices.

Details

PDF

Statistics

from
to
Export
Download Full History