首页--工业技术论文--自动化技术、计算机技术论文--计算技术、计算机技术论文--计算机的应用论文--信息处理(信息加工)论文--模式识别与装置论文

Automatic Fruit Recognition Based on DCNN for Commercial Source Trace System

ABSTRACT第5页
Abbreviations第8-9页
Chapter 1 Introduction第9-15页
    1.1 Motivation第11-12页
    1.2 Contribution of Study第12-13页
    1.3 Scope of study第13-15页
Chapter 2 Artificial Neural Networks第15-28页
    2.1 Perceptron第17-18页
    2.2 Multilayer perceptron第18-23页
        2.2.1 The training process第19-21页
        2.2.2 Backpropagation algorithm第21-23页
    2.3 Convolutional Neural Networks第23-24页
    2.4 Fruit Recognition and Detection Related Work第24-27页
    2.5 Issue and Challenges第27-28页
Chapter 3 DCNN Algorithm第28-45页
    3.1 Principal of Deep Convolutional Neural Network第28-30页
    3.2 Propose Model Architecture第30-37页
        3.2.1 Conv Net Layers第31页
        3.2.2 Convolutional Layer第31-34页
        3.2.3 Rectified Linear Unit(Re LU)第34-35页
        3.2.4 Pooling Layer第35-36页
        3.2.5 Fully Connected layer第36-37页
    3.3 CNN Algorithm & Back propagation Algorithm第37-39页
        3.3.1 Froward Path第37页
        3.3.2 Back Propagation第37-39页
    3.4 Training Using Backpropagation第39-41页
    3.5 Overfitting in Deep Neural Network第41-43页
        3.5.1 Over Fitting Detection第42页
        3.5.2 Methods to Avoid Over-fitting第42-43页
    3.6 CNN Design Principle第43-45页
Chapter 4 Experiments Results & Dataset第45-64页
    4.1 Dataset第45-53页
        4.1.1 Existing Fruit Dataset第46页
        4.1.2 Data Collection第46-50页
        4.1.3 Categorization Challenges第50-51页
        4.1.4 Data Augmentation第51-53页
    4.2 Software and Hardware第53页
    4.3 Methodology第53-57页
        4.3.1 Load The Dataset第54页
        4.3.2 Dimensionality Reduction and Gray Scaling第54-55页
        4.3.3 Data Preprocessing第55页
        4.3.4 Splitting our dataset第55-56页
        4.3.5 The Neural Network Architecture第56-57页
        4.3.6 Compile the Model第57页
        4.3.7 Train the model第57页
    4.4 Simulation Results第57-59页
        4.4.1 Adding Dropout into the Network第58-59页
    4.5 Predict Labels第59-60页
    4.6 Classification Report第60-61页
    4.7 Confusion Matrix第61-63页
    4.8 Comparison with other Algorithms第63-64页
Chapter 5 Transfer Learning with CNNs第64-68页
    5.1 Transfer Learning for New Class of fruit第65-67页
        5.1.1 Retrained the Output Dense Layer Only第65-66页
        5.1.2 Freeze the Weights of the first few layers第66-67页
    5.2 Experiment Results with Transfer Learning第67-68页
Chapter 6 Intra-Class Fruit Recognition System第68-77页
    6.1 Intra-Class Fruit Recognition System flow chart第68-69页
    6.2 Experiments and results第69页
    6.3 Apple Intra-Categorization第69-74页
        6.3.1 Dataset第69-71页
        6.3.2 Model Architecture第71页
        6.3.3 Experimental Process第71-72页
        6.3.4 Simulation Results and Analysis第72-73页
        6.3.5 Confusion Matrix第73-74页
    6.4 Kiwi Intra-Class Categorization第74-77页
        6.4.1 Experiment Process & Result Analysis第74页
        6.4.2 Dataset第74-75页
        6.4.3 Simulation Results, Classification table & Confusion Matrix第75-77页
Conclusion & Future Scope第77-78页
References第78-82页
Acknowledgement第82-83页
附件第83-84页
Publications(在学期间取得的科研成果)第84页

论文共84页,点击 下载论文
上一篇:基于RGB-D与深度卷积网络的地铁人流量统计算法研究
下一篇:基于半监督自步学习的跨任务深度网络应用于图像分类