Object Detectors for X-Ray Scanners

An automated object-recognition solution for X-ray imagery increases throughput and dramatically speeds up the screening process at transport hubs.

The need for manual inspection of thousands of items of luggage for illegal items creates an enormous workload. Thus, an intellectual detection system for X-ray baggage imagery is one of the most sought-after technologies. Despite the fact that obtaining X-ray baggage imagery is still a challenging task, the last few years of progress in computer vision and machine learning has brought this detection system to reality. Being one of pioneers in machine learning, BitRefine offers a powerful image-classification model plus R&D services in this innovative field.

Our object-detection system can work with objects such as:

• Firearms
• Stun guns
• Sprays
• Knives and scissors
• Screwdrivers
• Batteries
• Bottles

Our recognition technology is constantly improving with the addition of diverse representative data, training of the model, and optimizing of testing procedures. With experimentation on large data sets, the performance of detecting specific threats such as firearms already exceeds a 99% rate (for true positive). The extra benefit of our solution is the possibility of self-improving and self-learning, which allow continuously increasing performance of the model during runtime throughout the whole life cycle of the product. The percentage of misclassification of self-learning algorithms after a certain period drops down to values close to zero. At this moment, the system can be taken as a semi automated baggage-screening procedure. Link to BitRefine's X-ray object recognition software platform.