Material Recognition of Foreign Object Debris using Deep Learning

Abstract

Foreign Object Debris (FOD) is defined as one of the major problems in the airline maintenance industry, reducing the levels of safety. A foreign object which may result in causing serious damage to an airplane, including engine problems and personal safety risks. Therefore, it is critical to detect FOD in place to guarantee the safety of airplanes flying. FOD detection systems in the past lacked an effective method for automatic material recognition as well as high speed and accuracy in detecting materials. This paper proposes the FOD model using a variety of feature extraction approaches like Gray-level Co-occurrence Matrix (GLCM) and Linear Discriminant Analysis (LDA) to extract features and Deep Learning (DL) for classification. The data for this research was taken from the Shanghai Hongqiao International Airport runways. FOD has been done via utilizing a Convolutional Neural Network Classifier (CNN) with 27 layers. Those layers are distributed as follows: nine convolutional layers of type 1D; eight leaky ReLU layers; seven maxpooling 1D layers; two fully connected layers that are represented by the (Dense); and one flattening layer. The performance measures utilized in the system are precision, accuracy, F-score, and recall. The experimental results obtained after implementation and testing are of accuracy 99.8%, precision 100%, recall 100%, and F1-score is 100%. Experiments show that the proposed method has good performance in detection accuracy.