Vision-Based Activity Detection Inside Vehicles - Automotive In-Cabin Sensing

Explore a comprehensive conference talk on vision-based activity detection inside vehicles for automotive in-cabin sensing. Delve into the importance of activity detection in reducing road fatalities caused by distracted driving. Learn about the fusion of multiple signals required for robust detection, including body movement analysis, deep convolutional neural networks for object interaction, and established driver monitoring signals. Gain insights into the development process of an activity detection system, from data collection and annotation to integration on a System on Chip. Discover practical examples, tips, and tricks for developing various detection algorithms. The speaker, Mattias Ulmestrand, a technical developer at Smart Eye, shares his expertise in developing deep learning models for computer vision in embedded environments. Recorded at the 2023 GAIA Conference, this 25-minute talk covers topics such as DMS, modularity, object-in-hand detection, data collection, training, and a modular approach to activity detection.

Introduction
What is DMS
Modularity
Object in Hand
PreProduction
Data Collection
Training
Summary
Modular Approach
Questions