@TechReport{TR151,
  author      = "Anna Sebernegg",
  title       = "Benign Object Detection and Distractor Removal in 2D Baggage Scans",
  institution = "PRIP, TU Wien",
  number      = "PRIP-TR-151",
  year        = "2021",
  url         = "https://www.prip.tuwien.ac.at/pripfiles/trs/tr151.pdf",
  abstract    = "X-ray screening significantly impacts security applications such as baggage handling to help detect objects, especially threats such as explosives or weapons, within closed luggage otherwise not visible to the naked eye.  
                 However, the generated X-ray images are challenging to interpret due to the targets' weak visual signals in high background noise levels and the compact assembly of rotated and superimposed objects in bags.
                 The complexity of X-ray scans and the high intra-class variability of threats make appearance-based threat detection difficult for both human operators and automated systems.
                 Consequently, generic appearance-based threat detection systems are hardly available in practice, and baggage screening still depends highly on human operators. 
                 Nevertheless, further developments of automatic baggage inspection are desirable to support the visual search task of screeners.
                 This work proposes utilizing automatic benign object detection as a diagnostic aid, for instance, to remove distractors from the images through image inpainting. 
                 By reducing the number of distractive benign objects in the data, regions of interest could gain faster attention. 
                 The applied distractor removal methods successfully reduced visual saliency in regions of distractors and decreased the overall visual clutter of the X-ray scans.
                 ",
}