Researchers at University College London have developed a method based on deep learning for X-ray baggage to detect explosives

The development of phase-based technologies has accelerated the pace of X-ray imaging. Dark-field images are sensitive to inhomogeneities on a length scale below the system’s spatial resolution, and phase-contrast images are optimized to obtain a detailed view. A team of researchers from University College London, Niles Limited, and XPCI Technology Ltd. A new X-ray baggage technique to find trace levels of explosives, they show how the dark field produces texture specific to the material being imaged and how combining it with conventional attenuation improves the ability to distinguish between threat materials. They have also published their work in Nature Communications, which includes adapting a traditional X-ray detector and using a deep learning application to better detect dangerous chemicals in baggage.

In addition, their research demonstrates that persistent misunderstandings can be removed by using the different dark field energy dependencies and attenuation signals. In addition, two proof-of-concept experiments show that the dark field texture is suitable for identification using machine learning techniques. Performance suffered when identical methods were applied to datasets from which dark-field images had been omitted. Previous studies have shown that the type of material greatly influences the subtle curvatures produced when X-rays interact with it. The researchers aim to take advantage of these curvatures to build an accurate X-ray system.

The researchers’ first modification to existing X-ray equipment was to add a box containing the masks which are sheet metal with small holes punched through them. The purpose of the masks is to split an X-ray beam into several smaller beams. A deep learning AI app was then fed scan results from various objects with embedded explosive components. The goal was to educate the machine on how to perceive the appearance of the subtle bends of such materials. After the machine was trained, they tested its capabilities by scanning other items for built-in bombs. Under laboratory conditions, the researchers discovered that their devices achieved 100% accuracy.

The device succeeded in recognizing curvatures as small as one microradian, or one 20,000th of a degree. However, there is still room for further investigation because the research was done on a smaller scale. The overall results suggest that the use of deep neural networks and dark fields can be combined in applications outside of security. The team believes its method may be modified slightly for use in other applications, such as medicine, in addition to being useful for transportation security personnel. They think he might be trained to identify malignancies too small to be detected by standard testing equipment and to find subtle defects on the roofs of buildings or airplanes.

This Article is written as a research summary article by Marktechpost Staff based on the research paper 'Enhanced detection of threat materials by dark-field x-ray imaging combined with deep neural networks'. All Credit For This Research Goes To Researchers on This Project. Check out the paper and reference article.

Please Don't Forget To Join Our ML Subreddit


Khushbu Gupta is a Consultant Intern at MarktechPost. She is currently pursuing her Bachelor of Technology degree from the Indian Institute of Technology (IIT), Goa. She is passionate about the fields of machine learning, natural language processing, and web development. You enjoy learning more about the technical field by participating in many challenges.


Leave a Comment