General Attacks on Camera Systems

Cameras in ITS can detect traffic signs, delineation, or objects. These can be attacked in various ways: Traffic Sign Detection, can be fooled by placing fake traffic signs at improper locations or by hiding traffic signs with other shapes/colors to confuse the detection algorithms. Lane Detection can be confused by painting additional lines on the road or using different colors. Object Tracking is limited due to computational power or resolution. A denial of service can be caused by presenting too many objects to track. Deep Neural Networks (DNNs), which are used in camera software, can be easily fooled by images that are unrecognizable to humans but are recognized by DNNs with high confidence.