It was reported this week that an Austrian security researcher named Christian Haschek was able to build a homegrown system that can detect illegal images online. Haschek says his device cost only $300 to build and uses less power than a light bulb.
The device, which was trained to classify images, uses three Raspberry PIs powering two Intel Movidius sticks. Haschek also used an an open source algorithm called NSFW (Not Safe For Work) from Yahoo that can identify explicit material.
Prof Alan Woodward from Surrey University called Haschek’s project “encouraging”. He also lamented that the problem, which law enforcement agencies are struggling with, often take a lot of time and effort since it involves working with tech firms.