Driver behaviour detection is one of the most important parts of a vehicle driving safety assistant system. It helps prevent safety risks caused by human beings and reduces the occurrence of serious accidents significantly. Therefore, how to realize a mature driver behaviour detection application has drawn public attention. The information and connections between the driver, the car and the environment need to be considered to fully understand driver behaviour in realistic driving scenarios. This project mainly focuses on the driver behaviour application using computer vision in vehicles. This proposal aims to provide a general framework, requirements and evaluation methods for driver behaviour detection applications and to help companies and industries boost the development of driver behaviour detection on multimedia applications This Recommendation provides technical requirements and evaluation methods for AI-based driver behaviour detection application. This Recommendation could support the construction of such application by providing general user guidelines.