Law Enforcement to Stop Car Smuggling With Science
International trade hubs currently have a big problem: they are really, really big. So big, that is, that each shipping container can’t be visually checked before it is shipped out—a weakness that smugglers take advantage of to ship illegal cars, either by obscuring the vehicles or with other items to hide them from eyes. So, when you can’t accomplish the task with human eyes, what can you do?
Replace them with robot eyes, obviously. Research currently underway in the University College London’s Department of Computer Science has been developing and testing methods to have computers (specifically a convolutional neural network) examine X-rays of shipping containers and find automobiles. The network was very accurate, correctly identifying automobiles in every case, with very few false alarms—the computer was even able to identify cars intentionally obscured by other objects.
Apparently, this isn’t a new or surprising study, but makes for a great example of the possibilities of using deep learning image recognition (aka computers using X-rays to identify things) can potentially solve smuggling problems. You can almost see it now—entire cargo ships passing by scanners like oversized airplane luggage, computers finding the stolen vehicles bound for Africa before the tanker has the opportunity to leave port.
This could be used to identify any number of illegal items, with the study ending with the sentence, “Due to its generic nature, this deep learning scheme could likely be used to detect many classes of objects in complex X-ray imagery, even when only a modest dataset of examples is available.”