Threat Objects Detection in Airport using Machine Learning

Abstract

Airports operate under high demands and strive to satisfy many diverse, interrelated performance goals. Airport areas, such as security and safety also refer to detecting hazardous materials inside passengers’ luggage. This topic is independent to the size of the airport and presents a high importance. The detection of threat objects using X-ray luggage scan images is a big part of aviation security. Currently most screenings are still based on the manual detection of potential threat objects, and rely heavily on the human experts. Artificial intelligence is often used to describe machines that mimic cognitive human functions such as learning and problem solving. At this moment, the most interesting branch of artificial intelligence is automatic learning, allowing computers to learn to solve problems on their own. Machine learning (ML) made a series of discoveries that once seemed impossible, such as: recognizing the human voice, search engines, filtering out unsolicited messages, recognizing human face, medical diagnostics, creating realistic photos of people (using a set of photos, the algorithm creates new faces, practically inventing them), analyzing the stock market, recognizing objects, etc. The paper presents the implementation of a threat objects recognition system using the convolutional neural network and specialized libraries in ML such as OpenCV and Keras with TensorFlow. The topics presented are the development of the convolutional neural network, used data augmentation techniques, tests and detection rate using X-ray images.