Object Detection with Deep Learning Models: Principles and Applications

Cover
Half Title
Title Page
Copyright Page
Table of Contents
Editors
List of Contributors
Chapter 1: Introduction:: Deep Learning and Computer Vision
	1.1 Introduction to Deep Learning
		1.1.1 Deep Learning
		1.1.2 Machine Learning and Deep Learning
		1.1.3 Types of Networks in Deep Learning
			1.1.3.1 Connection Type of Networks
				1.1.3.1.1 Static Feedforward Networks
				1.1.3.1.2 Dynamic Feedback Neural Networks
			1.1.3.2 Topology-based Neural Networks
				1.1.3.2.1 Single-layer Neural Networks
				1.1.3.2.2 Multilayer Neural Networks
				1.1.3.2.3 Recurrent Neural Networks
			1.1.3.3 Learning Methods
				1.1.3.3.1 Supervised Learning
				1.1.3.3.2 Unsupervised Learning
				1.1.3.3.3 Reinforcement Learning
	1.2 Convolutional Neural Networks
		1.2.1 Description of Five Layers of General CNN Architecture
			1.2.1.1 Input Layer
			1.2.1.2 Convolutional Layer
			1.2.1.3 Pooling Layer
			1.2.1.4 Fully Connected Layers
			1.2.1.5 Output Layer
		1.2.2 Types of Architecture in CNN [ 9 ]
			1.2.2.1 LeNet-5
			1.2.2.2 AlexNet
			1.2.2.3 ZFNet
			1.2.2.4 GoogLeNet/Inception
			1.2.2.5 VGGNet
			1.2.2.6 ResNet
		1.2.3 Applications of Deep Learning
	1.3 Image Classification, Object Detection and Face Recognition
		1.3.1 Dataset Creation
		1.3.2 Data Preprocessing
		1.3.3 Image Classification
		1.3.4 Object Detection
		1.3.5 Face Recognition
	References
Chapter 2: Object Detection Frameworks and Services in Computer Vision
	2.1 Neural Networks (NNs) and Deep Neural Networks (DNNs)
		2.1.1 Neural Networks
		2.1.2 Single-Layer Perceptron (SLP)
		2.1.3 Multilayer Perceptron (MLP)
	2.2 Activation Functions
		2.2.1 Identity Function
		2.2.2 Sigmoid Function
		2.2.3 Softmax Function
		2.2.4 Tanh Function
		2.2.5 ReLU (Rectified Linear Unit) Function
	2.3 Loss Functions
	2.4 Convolutional Neural Networks
		2.4.1 CNN Architecture and its Components
	2.5 Image Classification Using CNN
		2.5.1 LeNet-5
		2.5.2 AlexNet
		2.5.3 VGGNet
		2.5.4 Inception and GoogLeNet
			2.5.4.1 Inception Module
		2.5.5 ResNet
			2.5.5.1 Residual Block
	2.6 Transfer Learning
		2.6.1 Need for Transfer Learning
		2.6.2 Transfer Learning Approaches
			2.6.2.1 Pre-trained Network as a Classifier
			2.6.2.2 Pre-trained Network as a Feature Extractor
			2.6.2.3 Fine Tuning
	2.7 Object Detection
		2.7.1 Object Localization
			2.7.1.1 Sliding Window Detection
			2.7.1.2 Bounding Box Prediction
		2.7.2 Components of Object Detection Frameworks
	2.8 Region-Based Convolutional Neural Networks (R-CNNs)
		2.8.1 R-CNN
		2.8.2 Fast R-CNN
			2.8.2.1 Components of Fast R-CNN
		2.8.3 Faster R-CNN
		2.8.4 YOLO Algorithm
		2.8.5 YOLOv1 Object Detection Model
		2.8.6 YOLO9000 Object Detection Model
		2.8.7 YOLOv3 Object Detection Model
	2.9 Computer Vision Application Areas
	References
Chapter 3: Real-Time Tracing and Alerting System for Vehicles and Children to Ensure Safety and Security, Using LabVIEW
	3.1 Introduction
	3.2 Scope of the Chapter
	3.3 System Requirements
		3.3.1 Hardware Requirements
		3.3.2 Software Requirements
	3.4 Real-Time Tracing and Alerting System Environment
		3.4.1 Image Acquisitions Module
		3.4.2 Object Identification Module
	3.5 System Architecture
	3.6 Outline of Vehicle-tracking Processing
	3.7 Overview of RFID
		3.7.1 Implementation of RFID
	3.8 Module
		3.8.1 Vehicle Tracking Using LabVIEW
		3.8.2 Student Tracking Using RFID
		3.8.3 Alerting System
	3.9 Conclusion
	References
Chapter 4: Mobile Application-based Assistive System for Visually Impaired People: A Hassle-Free Shopping Support System
	4.1 Introduction
		4.1.1 Cataracts
		4.1.2 Age-Related Macular Degeneration (AMD)
		4.1.3 Diabetic Retinopathy
		4.1.4 Glaucoma
	4.2 Related Works
		4.2.1 Item Identification
		4.2.2 Barcodes
	4.3 Proposed System
		4.3.1 Barcode Capture
		4.3.2 Barcode Detection
		4.3.3 Barcode Pre-processing
		4.3.4 Scan Distance
		4.3.5 User Feedback System
		4.3.6 Decoding of Barcode Image
		4.3.7 Fetching Product Specification
		4.3.8 Text-to-Speech Using Google TTS
	4.4 Experimental Result and Discussion
		4.4.1 Dataset Collection
	4.5 Conclusion and Future Work
	References
Chapter 5: Traffic Density and On-road Moving Object Detection Management, Using Video Processing
	5.1 Introduction
		5.1.1 Problem Definition
	5.2 Literature Survey
	5.3 Technical Concepts
		5.3.1 Image Processing
		5.3.2 Architecture Design
		5.3.3 Background Registration
		5.3.4 Color Identification
		5.3.5 Data Flow Diagram for Detection Modules
		5.3.6 Data Flow Diagram for Tracking Modules
		5.3.7 Data Flow Diagram for Counting Modules
	5.4 Proposed Methodology
		5.4.1 Proposed Method Steps
	5.5 Simulation and Result
	5.6 Conclusion
	References
Chapter 6: Automated Vehicle Number Plate Recognition System, Using Convolution Long Short-Term Memory Technique
	6.1 Introduction
	6.2 Literature Review
		6.2.1 Related Work Related to License Plate Recognition Technology
		6.2.2 Deep Learning-Based Work for Recognizing License Plates
	6.3 Methodology
		6.3.1 Convolutional LSTM
	6.4 Experiments
	6.5 Results
		6.5.1 Parameters for Evaluation
		6.5.2 Evaluation Metrics
		6.5.3 Comparison Evaluation Fusion Model With Baseline Models
	6.6 Conclusion
	References
Chapter 7: Deep Learning-Based Indian Vehicle Number Plate Detection and Recognition
	7.1 Introduction
	7.2 Literature Survey
	7.3 Proposed System
	7.4 Experimentation & Results
	References
Chapter 8: Smart Diabetes System Using CNN in Health Data Analytics
	8.1 Introduction
		8.1.1 What Is Big Data?
		8.1.2 Analytics in Big Data
		8.1.3 Healthcare – Big Data Analytics
			8.1.3.1 Challenges
			8.1.3.2 Developing Complexity of Healthcare Information
		8.1.4 Big Data Framework
			8.1.4.1 Cloud and Big Data
			8.1.4.2 Open Source Arrangements for Big Data Information in Healthcare
	8.2 Problem Identification
	8.3 Proposed Solution
		8.3.1 Smart Diabetes System
		8.3.2 Objectives
			8.3.2.1 Personalized Information Investigation Demonstration for Smart Diabetes
	8.4 5G Smart Diabetes Model – Technologies
		8.4.1 Fifth Generation Mobile Networks
		8.4.2 Machine Learning Techniques
			8.4.2.1 AI Applications in Healthcare
				8.4.2.1.1 Discrete occasion simulation
				8.4.2.1.2 Free-text doctor notes
		8.4.3 How Convolution Neural Network Applies Here
		8.4.4 Medical Big Data
			8.4.4.1 Diabetes 1.0
			8.4.4.2 Diabetes 2.0
		8.4.5 Social Networking
		8.4.6 Smart Clothing
	8.5 Smart Diabetes Architecture
		8.5.1 Smart Diabetes Design
		8.5.2 Detection
		8.5.3 Personalized Determination
		8.5.4 Information Sharing
			8.5.4.1 Social Space
			8.5.4.2 Information Space
			8.5.4.3 How Can Social and Information Space Be Combined?
		8.5.5 System Sensor Architecture
			8.5.5.1 How Does the Continuous Glucose Monitor (CGM) Work?
			8.5.5.2 Phenomenal Highlights of a CGM
			8.5.5.3 Unprecedented Necessities Required to Utilize a CGM
			8.5.5.4 Who Can Utilize a CGM?
			8.5.5.5 What Are the Benefits of a CGM?
			8.5.5.6 What Are the Constraints of a CGM?
			8.5.5.7 What Could Put Everything in Order for an Artificial Pancreas?
	8.6 5G Smart Diabetes System Test Bed & Result
		8.6.1 Information Collection from a Healing Community
		8.6.2 Diet
		8.6.3 Exercise
		8.6.4 Sharing Information
		8.6.5 The Test Bed of Machine Learning Calculations
		8.6.6 Results
	8.7 Conclusion
	References
Chapter 9: Independent Automobile Intelligent Motion Controller and Redirection, Using a Deep Learning System
	9.1 Introduction
	9.2 Related Work
	9.3 Existing System
	9.4 Proposed Method: Two-tier Approach for AI Transportation Traffic Flow Administration
		9.4.1 Optimization of Traffic Lights
	9.5 Smart Redirected Path Use
	9.6 Discussions and Results
		9.6.1 First Layer
		9.6.2 Second Layer
		9.6.3 Third Layer
		9.6.4 Fourth Layer
		9.6.5 Fifth Layer
	9.7 Conclusion
	9.8 Acknowledgment
	References
Chapter 10: Deep Learning Solutions for Pest Detection
	10.1 Introduction
		10.1.1 Object Detection
		10.1.2 Deep Object Detection
			10.1.2.1 Types of Deep Object Detection
		10.1.3 Challenges in Object Detection
	10.2 Advances in Agriculture
		10.2.1 Smart Farming
		10.2.2 Deep Learning in Agriculture
		10.2.3 Automatic Pest Detection
		10.2.4 Challenges in Automatic Pest Detection
			10.2.4.1 Extrinsic Factors
			10.2.4.2 Intrinsic Factors
			10.2.4.3 Big Data Availability for Deep Detection
	10.3 Novel Smart Intelligent System for Paddy Pest Detection
		10.3.1 Related Work
		10.3.2 Training Phase
			10.3.2.1 Dataset Description
			10.3.2.2 Classes Used in the Proposed Method
		10.3.3 EfficientDet
		10.3.4 Server Framework
		10.3.5 Mobile Service Framework
		10.3.6 Performance Metrics for Evaluation
			10.3.6.1 Precision and Recall
			10.3.6.2 Average Precision (AP)
			10.3.6.3 Mean average Precision (mAP)
			10.3.6.4 Precision-Recall Curve
			10.3.6.5 Inference Speed
			10.3.6.6 Service Time for User
	10.4 Conclusion
	References
Chapter 11: Deep Learning Solutions for Pest Identification in Agriculture
	11.1 Introduction
	11.2 Existing Literature
		11.2.1 Disease Detection
		11.2.2 Land Cover Identification
		11.2.3 Classification of Plants
		11.2.4 Precision Livestock Farming
		11.2.5 Pest Recognition
	11.3 Background Details
		11.3.1 Deep Learning
		11.3.2 Motivation of this Study
		11.3.3 Contribution
			11.3.3.1 Similarity of Different Types of Plant Disease
			11.3.3.2 Steps Involved in Plant Disease Detection
			11.3.3.3 Deep Learning in Tomato Diseases
			11.3.3.4 Deep Learning in Potato Diseases
			11.3.3.5 Deep Learning in Apple Diseases
			11.3.3.6 Deep Learning Approaches for High Spectral Images in Agricultural Field
	11.4 Conclusion
	References
Chapter 12: A Complete Framework for LULC Classification of Madurai Remote Sensing Images with Deep Learning-based Fusion Technique
	12.1 Introduction
	12.2 Related Work
		12.2.1 Image Fusion
		12.2.2 Process of Feature Extraction
		12.2.3 Feature Selection
		12.2.4 Classification of Images
	12.3 Problem Statement
	12.4 Proposed Work
		12.4.1 System Overview
		12.4.2 Image Fusion
		12.4.3 Feature Extraction
			12.4.3.1 Deep Features
				12.4.3.1.1 Convolutional Layer
				12.4.3.1.2 Pooling Layer
			12.4.3.2 Gray Level Co-occurrence Matrix (GLCM)
			12.4.3.3 Hu Invariant Moments
			12.4.3.4 Color Moments
		12.4.4 Feature Selection
			12.4.4.1 Ranking Procedure
			12.4.4.2 Reconstruction Error (RE) Measure
		12.4.5 Image Classification
			12.4.5.1 Classification Based on BP Algorithm
			12.4.5.2 Classification Based on k-nearest Neighbor
			12.4.5.3 Classification Based on Naive Bayes
			12.4.5.4 Combined Classifier System (CCS)
	12.5 Experimental Results and Discussion
		12.5.1 Description of Dataset
		12.5.2 Results of the Proposed System
			12.5.2.1 Evaluation for Image Fusion
			12.5.2.2 Evaluation for Classification
		12.5.3 Discussions about the Proposed System
	12.6 Conclusion
	References
Chapter 13: Human Behavioral Identifiers: A Detailed Discussion
	13.1 Introduction to Biometric Technology
	13.2 Historical Outline
	13.3 The Basic Characteristics
		13.3.1 Collectability
		13.3.2 Circumvention
		13.3.3 Distinctiveness
	13.4 Biometric Types
		13.4.1 Fingerprints
		13.4.2 Photo and Video
		13.4.3 Speech
		13.4.4 Signature
		13.4.5 DNA
	13.5 Behavioral Identifiers
		13.5.1 Inputting Forms
		13.5.2 Physical Movements
		13.5.3 Navigation Forms
		13.5.4 Engagement Patterns
	13.6 Applications
		13.6.1 Financial Sector
		13.6.2 Security
		13.6.3 Mobile Application Domain
		13.6.4 Justice, Law and Enforcement Applications
		13.6.5 Public Services Applications
			13.6.5.1 Healthcare
			13.6.5.2 Border Control and Airports
		13.6.6 Eye Movement Tracking Applications
			13.6.6.1 Aviation
			13.6.6.2 Automotive Industry
			13.6.6.3 Screen Navigation
	13.7 The Rise of Static Biometric Authentication through Physical Characteristics
	13.8 Behavioral Biometrics in Today’s Digital World
	13.9 Analyzing the Patterns in Human Activity
		13.9.1 Physical Movements
		13.9.2 Voice Biometrics
		13.9.3 Device-based Gestures
	13.10 Emerging Technologies in Behavioral Biometrics
		13.10.1 Human Behavioral Patterns
		13.10.2 Sensors
	13.11 Machine Learning/Deep Learning
		13.11.1 How it Works
	13.12 Behavioral Biometrics Examples
		13.12.1 Compromised Credentials
		13.12.2 Account Details/Password Sharing
		13.12.3 User Substitution
		13.12.4 Remote Access Trojans
		13.12.5 Insider Threats
		13.12.6 USB Rubber Ducky Attacks
		13.12.7 Phishing Attacks
		13.12.8 Uncertain Attribution
		13.12.9 User/Client Carelessness
		13.12.10 Identity Fraud
		13.12.11 License Mismanagement
	13.13 Merits and Demerits
	13.14 Future of Behavioral Biometrics
	References
Index