Securing Camera Based Active Driver Monitoring System from Video Forgery Attacks Using Deep Learning 2022-01-0115
The numerous superiorities of autonomous vehicles in terms of safety, driving experience, and comfort against the traditional driving favor them in the wide adoption across the modern automotive sector. Driver Monitoring System (DMS) is one of the high Automotive Safety Integrity Level (ASIL) specified driver assistance functionalities, which assists the driver continuously as a part of Active Safety system. The fused vision-related functionalities of the camera based active DMS such as Distraction, Drowsiness, and Emotion detection systems monitors the driver’s In-vehicle Activities, Eye movements, and Facial expressions respectively to awake the driver with constant alerts under undesired conditions to achieve a right driving attention. The presence of the modern day DMS in the In-Vehicle-Infotainment Digital Cockpits exposes the critical DMS into a wide range of cyber-attacks either locally or remotely, which in turn causes the malfunctioning of the active safety driver assistance system. This malfunctioning is due to integrity compromisations through cyberattacks, where attackers forge the incoming driver video data on the compromised infotainment system, which has external world connectivity. The tampering of incoming video content which contains driver activities, eye movements, and facial expressions at pixel level, block level or scene results in false interpretation by the distraction, drowsiness, and emotion detection modules of DMS respectively. The false interpretation of the driver states by camera based behavioral active DMS causes numerous false positive and false negative driver alerts due to cyber-attacks, which are undesired. Hence, here a deep learning based digital forensic approach is proposed to detect the forgery attacks on the DMS. The attacker’s successful launch of forgery attacks on the driver monitoring video information is detected intelligently using deep learning techniques before passing the video input for DMS modules such as Distraction, Drowsiness, and Emotion detection. The proposed approach involves the detection of splicing and copy-move object forgery attacks on the front camera feeds of the DMS in real-time using convolutional neural networks. The proposed system is responsible for validating the authenticity of the incoming real-time camera video frames prior to Distraction, Drowsiness, and Emotion detections and generating alerts. Hence, the proposed system successfully detects the video forgery attacks, and provides only the authentic driver monitoring alerts by enhancing the trust on driver monitoring systems from cyber-attacks by ensuring the safety and system integrity.