The number of passengers traveling by air is continuously on the rise. According to the World Bank, ~3.7 billion passengers travel by air globally (with ~7% increase every year). The UAE is witnessing an extraordinary growth in both land and air passenger traffic. In Abu Dhabi’s airport for instance, the passenger traffic is growing at a rate of 19% annually. Such an expansion, together with the cosmopolitan population in the UAE, safety and security concerns in today’s world and middle eastern conjuncture, in particular, require fast and foolproof security measures. To ensure safety, the security staff at ports manually inspect incoming and outgoing passenger luggage, packages, and containers with x-ray scanners. The screening process is resource intensive and requires constant attention of human experts during monitoring. This introduces an additional risk of human error caused by fatigued work-schedule, difficulty in catching contraband items, requirement of quick decision making, or simply due to a less experienced worker. In the case of passenger luggage, research shows that threat detection performance of human experts is only about 80–90%. Furthermore, human inspectors can only inspect a fraction of the whole incoming international mail. An intelligent, automated system for the detection of contraband items in X-ray images would bring significant contribution in mitigating the aforementioned issues. To that end, there are several challenges to overcome:
Based on the above, a practical contraband detection system for security and regulation applications that is accurate and reliable is therefore far from realization and is an open research problem. This project aims to find novel solutions to address the crucial limitations of the existing systems.