Paper Analysis - Driver Behavior and Interaction with Automation

Paper - https://arxiv.org/pdf/1711.06976.pdf

Key Objectives Studied and Techniques Identified for those key features

• Capture Driver State (Fine-grained face recognition, facial expression recognition, detect emotional stress)
• Driver body position (Driver Pose, Driving Behavior, Driver Alertness monitoring)
• Driver Scene Perception (Places / Landscapes
• Vehicle State - Driving Behavior, Steering Commands

Frame Details

• Three or four Logitech C920 webcams record at a resolution of 1280x720 at 30 frames per second within the car

Software

• Cam2hd is a program written in C that opens and records all camera data
• Lighthouse is a python script that sends information about each trip to Homebase. Information sent includes timing information for the trip, GPS data, power consumption, temperature, and available external drive space
• RIDER Database - A PostgreSQL database is used to store all incoming trip information, as well as to house information about all trips offloaded to a storage server

A Good start to understand/reuse this approach/hardware recommendation for large-scale video analytics projects

Happy Learning!!!