Screenings for Forbidden Objects
In the realm of security and computer vision, accessing large datasets of X-ray images specifically depicting prohibited items such as guns, knives, and other objects at security checkpoints can be challenging. This is due to security and privacy concerns that prevent the public availability of such datasets.
However, there are several approaches to access these valuable resources:
- Academic and Research Datasets: Some datasets, particularly those focused on baggage screening and computer vision, are shared within the research community. Examples include the GDXray dataset, which features industrial X-ray images including defects and some security objects, or synthetic X-ray datasets generated for AI training. Access to these datasets typically requires research credentials or collaboration.
- Collaborations with Security Agencies or Institutions: Organizations like airport security authorities or customs departments develop and hold large proprietary datasets. Access to these datasets often requires official partnerships, security clearances, or involvement in approved research projects.
- Specialized Research Programs and Competitions: Occasionally, datasets are released in conjunction with international challenges on the detection of prohibited items in X-ray images. These are limited-time and controlled-access releases.
- Commercial or Industrial Providers: Companies developing screening technology, such as AI for X-ray image detection, may offer licensed datasets. However, these are rarely public.
A thorough search does not reveal any currently public, largest dataset of X-ray images featuring prohibited items. The discussions revolve around regulations on screening, the use of image retrieval and machine learning in security contexts, but no direct mention of a publicly sharable large X-ray dataset containing guns, knives, and other threat items is found.
Recent technological advancements, such as the one at Universiti Teknologi Malaysia, combine machine learning with an image database of prohibited items used in radiation and cargo scanning for security purposes. However, this system appears proprietary and specific to that institution, rather than a general dataset available for download or research access.
In conclusion, no freely accessible largest dataset of X-ray images of prohibited items at checkpoints is currently known or openly available. Access would likely require research collaboration or institutional authorization with security agencies or vendors that develop and control such data. For research purposes, contacting security labs or exploring existing synthetic/academic datasets in the scientific literature is recommended.
Interestingly, researchers at State Key Laboratory of Computer Science and Beihang University in China, and JD Finance America Corporation in the United States, have recently released a dataset of x-ray images of items commonly prohibited at security checkpoints. The origin of one such image is credited to Flickr user Thomas Hawk, and the dataset contains a substantial 47,677 x-ray images. This development may provide a valuable resource for the research community in the future.
- To expand the research on AI and security, researchers could explore the recently released dataset by the State Key Laboratory of Computer Science at Beihang University in China and JD Finance America Corporation, which contains 47,677 x-ray images of items commonly prohibited at security checkpoints, including some originating from Flickr user Thomas Hawk.
- Despite the lack of a free, largest dataset of X-ray images featuring prohibited items at checkpoints, researchers can access academic and research datasets like the GDXray dataset, which includes industrial X-ray images featuring security objects, or synthetic X-ray datasets generated for AI training, typically requiring research credentials or collaboration.
