In this paper, we describe how the electronic rolling shutter in CMOS image\nsensors can be exploited using a bright, modulated light source (e.g., an\ninexpensive, off-the-shelf laser), to inject fine-grained image disruptions. We\ndemonstrate the attack on seven different CMOS cameras, ranging from cheap IoT\nto semi-professional surveillance cameras, to highlight the wide applicability\nof the rolling shutter attack. We model the fundamental factors affecting a\nrolling shutter attack in an uncontrolled setting. We then perform an\nexhaustive evaluation of the attack's effect on the task of object detection,\ninvestigating the effect of attack parameters. We validate our model against\nempirical data collected on two separate cameras, showing that by simply using\ninformation from the camera's datasheet the adversary can accurately predict\nthe injected distortion size and optimize their attack accordingly. We find\nthat an adversary can hide up to 75% of objects perceived by state-of-the-art\ndetectors by selecting appropriate attack parameters. We also investigate the\nstealthiness of the attack in comparison to a na\\"{i}ve camera blinding attack,\nshowing that common image distortion metrics can not detect the attack\npresence. Therefore, we present a new, accurate and lightweight enhancement to\nthe backbone network of an object detector to recognize rolling shutter\nattacks. Overall, our results indicate that rolling shutter attacks can\nsubstantially reduce the performance and reliability of vision-based\nintelligent systems.\n