使用Plant Village进行深度学习和特征提取的植物病害检测
发布时间：2021-10-30 19:43
本文利用深度学习和特征提取技术解决植物病害检测问题。所有测试和实验都是使用开源数据集Plant Village进行的。本文的主要工作是实现三种不同的深度学习模型,即Resnet 50,Google Net和VGG16,并找出其中最适合解决分类问题的网络模型。众所周知,世界人口约为70亿,农作物疾病是世界粮食供应的关键问题,而超过90%的人无法使用能够识别和解决植物病问题的工具或功能。如今,我们生活在一个由大规模技术,主要网络覆盖,高端智能手机以及人工智能的发现和改进所主导的世界中。将高端智能手机和基于深度学习的计算机视觉相结合成为了可能。人们将其定义为“智能手机辅助疾病诊断”。在深度学习领域,学者们已经训练了多种架构模型,其中一些模型的性能达到了99.53%以上。先前的研究是对每种模型分别进行的,每个模型都产生自己的结果。但是,在我们的研究中,我们使用最新的技术将三个先前测试过的深度学习模型（Resnet50,Google Net,VGG16）和两个分类器（SVM和KNN）组合在一起,以便比较获得的结果并找出哪种模型能够更准确且更好地解决植物病害分类问题。在本文中,我们解决了这个问题。...
【文章来源】：大连理工大学辽宁省 211工程院校 985工程院校 教育部直属院校
【文章页数】：59 页
【学位级别】：硕士
【文章目录】：
摘要
Abstract
1 Introduction
1.1 Overview
1.2 Definition of Plant Diseases Detection
1.3 Why is it a real world problem?
1.4 Objectives and Contributions
1.5 Thesis Structure
2 Related Work
2.1 Previous Works on Plant diseases detection
3 Background
3.1 General Neural Networks
3.1.1 Convolution
3.1.2 Max Pooling
3.2 Recurrent Neural Networks
3.2.1 Image processing in smart agriculture
3.2.2 Pre-processing
3.2.3 Segmentation
3.2.4 Crop detection
3.2.5 Use of tracking algorithms
3.2.6 Plant disease classification using deep learning
3.3 Overfitting
4 Proposed Method
4.1 Introduction
4.2 System Architecture
4.2.1 Deep feature extraction architecture
4.2.2 Transfer Learning architecture
4.3 Data collection and dataset preparation
4.3.1 Data Collection
4.3.2 Dataset preparation
4.4 Data preprocessing
4.5 Pre-trained CNN models and deep learning networks
4.5.1 VGG16 network
4.5.2 Google net network
4.5.3 Resnet50 network
4.6 Classification algorithm
4.6.1 Support vector machine( SVM)
4.6.2 K-Nearest neighbor(KNN)
4.7 Performance and evaluation metrics
4.8 Equipment’s configuration and libraries
5 Result and discussion
5.1 Feature extraction results with Resnet50,Google Net and VGG
5.2 Deep learning results based on Resnet50,Google Net and VGG16
5.3 Results based on traditional shallow Networks
5.4 Discussion
6 Conclusion and Future Work
References
Research Projects and Publications in Master Study
Acknowledgements
本文编号：3467314
【文章来源】：大连理工大学辽宁省 211工程院校 985工程院校 教育部直属院校
【文章页数】：59 页
【学位级别】：硕士
【文章目录】：
摘要
Abstract
1 Introduction
1.1 Overview
1.2 Definition of Plant Diseases Detection
1.3 Why is it a real world problem?
1.4 Objectives and Contributions
1.5 Thesis Structure
2 Related Work
2.1 Previous Works on Plant diseases detection
3 Background
3.1 General Neural Networks
3.1.1 Convolution
3.1.2 Max Pooling
3.2 Recurrent Neural Networks
3.2.1 Image processing in smart agriculture
3.2.2 Pre-processing
3.2.3 Segmentation
3.2.4 Crop detection
3.2.5 Use of tracking algorithms
3.2.6 Plant disease classification using deep learning
3.3 Overfitting
4 Proposed Method
4.1 Introduction
4.2 System Architecture
4.2.1 Deep feature extraction architecture
4.2.2 Transfer Learning architecture
4.3 Data collection and dataset preparation
4.3.1 Data Collection
4.3.2 Dataset preparation
4.4 Data preprocessing
4.5 Pre-trained CNN models and deep learning networks
4.5.1 VGG16 network
4.5.2 Google net network
4.5.3 Resnet50 network
4.6 Classification algorithm
4.6.1 Support vector machine( SVM)
4.6.2 K-Nearest neighbor(KNN)
4.7 Performance and evaluation metrics
4.8 Equipment’s configuration and libraries
5 Result and discussion
5.1 Feature extraction results with Resnet50,Google Net and VGG
5.2 Deep learning results based on Resnet50,Google Net and VGG16
5.3 Results based on traditional shallow Networks
5.4 Discussion
6 Conclusion and Future Work
References
Research Projects and Publications in Master Study
Acknowledgements
本文编号：3467314
本文链接：https://www.wllwen.com/projectlw/swxlw/3467314.html
