What is in-cabin monitoring systems’ ultimate goal? That’s right, you guessed it: Safety.

If there is one thing that perception systems need when safety is at stake, that is high levels of accuracy. Many in-cabin monitoring systems are camera-based, deep learning-powered perception systems to detect all kinds of objects and situations that can happen inside a vehicle under different circumstances. Accuracy, in this context, means the confidence the deep learning models give you in detecting what you need in each case. The metrics typically used to measure that confidence are the precision and recall of the model when facing a test dataset.

Highlights from the session:

• What can we do to integrate data engineering in the perception system fine-tuning process?
• High data accuracy is mandatory when developing an in-cabin monitoring AI model
• Bridging the sim2real gap