目录
摘 要 Ⅰ
ABSTRACT Ⅱ
第一章 绪论 1
1 研究目的和意义 1
2 国内外研究现状 1
2.1国内研究现状 1
2.2 国外研究现状 2
3 研究内容和论文结构排版 3
3.1 研究内容 3
3.2论文结构排版 3
第二章 材料与方法 4
1数据 4
1.1 数据采集 4
1.1 数据处理 4
2方法 5
2.1技术简介 5
2.2 YOLO算法分析 7
2.3 YOLO算法优势 9
第三章 结果与讨论 12
1训练数据 12
2 结果分析 14
第四章 结论与展望 16
1 结论 16
2 展望 16
参考文献 18
致 谢 20
基于视觉技术和深度学习的鸡热应激行为识别研究
摘 要
随着“动物福利养殖”科学概念在全球的逐渐普及，环境质量逐渐成为制约鸡群健康福利及生产性能的重要因素之一。本项目以舍内平家禽黄羽鸡为主要研究对象，利用机算计视觉技术和物联网等信息化技术，构建了一套家禽养殖的智能化监控管理系统，主要有温湿度监测系统、氨气监测系统、内外循环的通风系统、视频监测系统，并且利用深度学习中经典的YOLO算法来进行鸡的个体识别与追踪、行为的分类识别以及体况、疾病检测等方面，通过自动化的方式进行信息监测，对鸡舍处于群体状态时的信息进行采集、监测畜禽的四种行为：饮水，吃食，振翅和其他行为等相关信息。本文在实验过程中制作了上万张数据集，利用深度学习算法对于鸡的行为进行识别，及时发现鸡的不良行为，使家禽信息的实时监测更有利于精准化绿色养殖。养殖者根据自动化系统能够及时调整鸡舍环境，保证鸡群的健康生活环境，同时也增加了鸡肉的产量，提高生产效益，增加收入；对鸡群身体状况下降时进行早期报警，可以提醒养殖者对可能发生的事故进行排查，预防大规模疫病的爆发，保证禽肉禽蛋的食品安全。
关键字：深度学习；视觉技术；鸡行为分析；物体识别 *51今日免费论文网|www.51jrft.com +Q: ^351916072# 

RESEARCH OF CHICKEN HEAT STRESS BEHAVIOR BASED ON VISUAL TECHNOLOGY AND DEEP LEARNING
ABSTRACT
With the popularization of the scientific concept of "animal welfare breeding", environmental quality is an important factor restricting the health, welfare and production performance of chicken flocks. Taking yellowfeathered chickens in the house as the research object, this project adopts machine vision technology and Internet of Things technology to build an intelligent monitoring system for poultry breeding, including temperature and humidity monitoring system, internal and external circulation ventilation system, and video monitoring system. In addition, the classic YOLO algorithm in deep learning is also used. Under a complex background, denoising and preprocessing are used to identify and track individual chickens, classify and identify behaviors, detect body conditions and diseases, etc. Through noncontact information monitoring, the four behaviors of collecting information and monitoring livestock and poultry when the chicken house is in a group state are as follows: we have produced tens of thousands of data sets for drinking water, eating, wing flutter and other behaviors, and used the deep learning algorithm to identify the behavior of chickens and find out the bad behavior of chickens in time, so that the realtime monitoring of poultry information is more conducive to accurate and green farming. The farmers can breed environment in time to ensure the breeding welfare, and also increase the production of chicken and improve the production efficiency. The early warning of the decline of the health condition of the chickens can remind the farmers to investigate the possible diseases, prevent the outbreak of largescale diseases, and ensure the food safety of poultry meat and eggs.

原文链接：http://www.jxszl.com/dzxx/dzdq/609083.html