Deep Learning based Framework for Automatic Damage Detection in Aircraft Engine Borescope Inspection

  title={Deep Learning based Framework for Automatic Damage Detection in Aircraft Engine Borescope Inspection},
  author={Zejiang Shen and Xili Wan and Feng Ye and Xinjie Guan and S. Liu},
  journal={2019 International Conference on Computing, Networking and Communications (ICNC)},
  • Zejiang Shen, Xili Wan, S. Liu
  • Published 1 February 2019
  • Computer Science
  • 2019 International Conference on Computing, Networking and Communications (ICNC)
To ensure high safety in civil aviation, borescope inspection has been widely applied in early damage detection of aircraft engines. Current manual damage inspection on borescope images inevitably results in low efficiency for engine status inspection. Traditional recognition methods are inefficient for damage detection due to complicated and noisy scenarios inside them. In this paper, a deep learning based framework is proposed which utilizes the state-of-the-art algorithm called Fully… 

Figures and Tables from this paper

Corrosion Detection on Aircraft Fuselage with Multi-teacher Knowledge Distillation

This work demonstrates that the novel DL ensembling scheme, i.e., multi-teacher/single-student knowledge distillation architecture, allows for 100% detection of the images representing the ”moderate corrosion” class on the test set, and shows that the proposed ensemble classifier, when used for the whole dataset with images representing various stages of corrosion, yields improved accuracy in comparison to the baseline single ResNet50 neural network.

Automatic Borescope Damage Assessments for Gas Turbine Blades via Deep Learning

This work proposes an automatic workflow based on deep learning for detecting damage present on rotor blades using borescope videos, and shows that damage statistics can be presented for each blade in a blade row separately, and demonstrates the workflow on two Borescope videos.

Automated Defect Detection and Decision-Support in Gas Turbine Blade Inspection

The work shows that image-processing approaches have potential value as a method for detecting defects in small data sets and identifies which viewing perspectives are more favourable for automated detection, namely, those that are perpendicular to the blade surface.

Pose Estimation and Damage Characterization of Turbine Blades during Inspection Cycles and Component-Protective Disassembly Processes

Inspection in confined spaces and difficult-to-access machines is a challenging quality assurance task and particularly difficult to quantify and automate. Using the example of aero engine

Object Dimension Estimation for Remote Visual Inspection in Borescope Systems

A single camera measurement solution that enables the size estimation of viewed objects is proposed, which computes and overlays a scaled grid of known spacing value over the screen view, enabling the human inspector to estimate the size of the objects in view.

YOLOv3-Lite: A Lightweight Crack Detection Network for Aircraft Structure Based on Depthwise Separable Convolutions

YOLOv3-Lite can reach state-of-the-art performance and is a fast and accurate crack detection method, which can be used on aircraft structure such as fuselage or engine blades.



Deep Learning‐Based Crack Damage Detection Using Convolutional Neural Networks

This article proposes a vision‐based method using a deep architecture of convolutional neural networks (CNNs) for detecting concrete cracks without calculating the defect features, and shows quite better performances and can indeed find concrete cracks in realistic situations.

Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification

This work proposes a Parametric Rectified Linear Unit (PReLU) that generalizes the traditional rectified unit and derives a robust initialization method that particularly considers the rectifier nonlinearities.

Deep Residual Learning for Image Recognition

This work presents a residual learning framework to ease the training of networks that are substantially deeper than those used previously, and provides comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth.

ImageNet classification with deep convolutional neural networks

A large, deep convolutional neural network was trained to classify the 1.2 million high-resolution images in the ImageNet LSVRC-2010 contest into the 1000 different classes and employed a recently developed regularization method called "dropout" that proved to be very effective.

SSD: Single Shot MultiBox Detector

The approach, named SSD, discretizes the output space of bounding boxes into a set of default boxes over different aspect ratios and scales per feature map location, which makes SSD easy to train and straightforward to integrate into systems that require a detection component.

A stereovision-based crack width detection approach for concrete surface assessment

To quantitatively evaluate crack width of concrete structures surface, this paper presents a stereovision-based crack width detection method. Compared with the traditional visual inspection with

Accurate Pulmonary Nodule Detection in Computed Tomography Images Using Deep Convolutional Neural Networks

This paper introduces a deconvolutional structure to Faster Region-based Convolutional Neural Network (Faster R-CNN) for candidate detection on axial slices and proposes a novel pulmonary nodule detection approach based on DCNNs.

Very Deep Convolutional Networks for Large-Scale Image Recognition

This work investigates the effect of the convolutional network depth on its accuracy in the large-scale image recognition setting using an architecture with very small convolution filters, which shows that a significant improvement on the prior-art configurations can be achieved by pushing the depth to 16-19 weight layers.

Rethinking the Inception Architecture for Computer Vision

This work is exploring ways to scale up networks in ways that aim at utilizing the added computation as efficiently as possible by suitably factorized convolutions and aggressive regularization.