Author Description

Front Cover -- IoT and Spacecraft Informatics -- Copyright Page -- Dedication -- Contents -- List of contributors -- About the editors -- Foreword -- Preface -- Acknowledgment -- 1 Artificial intelligence approach for aerospace defect detection using single-shot multibox detector network in phased arr... -- 1.1 Introduction -- 1.1.1 Ultrasonic inspection in aircraft -- 1.1.2 Autonomous inspection -- 1.2 Literature review -- 1.2.1 Composite material for the aerospace industry -- 1.2.2 Defects on composite materials -- 1.2.3 Defect inspection of composite materials -- 1.3 Defect detection algorithm -- 1.3.1 R-convolutional neural network -- 1.3.2 You only look once -- 1.3.3 Single-shot mulibox detector -- 1.3.4 Single-shot mulibox detector versus you only look once -- 1.3.5 Convolutional neural network-based object detection in nondestructive testing -- 1.4 Deployment of defect detection -- 1.4.1 Setting up of the deep learning environment -- 1.4.1.1 NVidia Tensorflow Object Detection API -- 1.4.1.2 TensorRT -- 1.4.1.3 OpenCV -- 1.4.2 Model training -- 1.4.3 Deployment in NVidia jetson TX2 -- 1.4.3.1 Program structure -- 1.4.3.2 OpenCV -- 1.4.3.3 MQTT -- 1.4.4 Validation -- 1.5 Implementation -- 1.5.1 Dataset preparation -- 1.5.2 Defect scanning -- 1.5.3 Image augmentation -- 1.5.4 Image annotation -- 1.6 Results -- 1.6.1 Loss -- 1.6.1.1 Classification loss and localization loss -- 1.6.1.2 Network configuration comparison and improvement -- 1.6.2 Validation of the defect detection system -- 1.6.2.1 Validation test sets -- 1.6.2.2 Manual labeling -- 1.6.2.3 Preliminary result of system and improvement -- 1.6.2.4 Automatic inspection -- 1.6.2.5 Comparison between automatic and manual inspection -- 1.7 Conclusions -- Acknowledgment -- References -- 2 Classifying asteroid spectra by data-driven machine learning model -- 2.1 Introduction.