This PhD project focuses on in-cabin scene analysis by using deep learning methodologies. Advanced safety is an important feature of autonomous vehicles. With the development of driving safety, vehicle interior scene recognition (including human body and head posture, motions, etc.) is required to provide a safe and comfortable driving environment. In this project, we focus on designing a vehicle in-cabin monitoring system to monitor and track drivers, as well as all passengers within the vehicle. We will use deep learning methods for object detection and semantic segmentation in vehicles, achieve real-time 3D human shape cabin modeling and annotate a 3D vehicle in-cabin scene database for further scientific use. The output of this PhD research will provide a monitoring system with automatic safety, comfort and surveillance functions to fill the gaps in the in-cabin scene analysis.