Deep Learning and Index Fusion-Based Driving Monitoring System
摘要
This empirical study presents an embedded system for real-time driving monitoring. This proposal is based on three convolutional deep learning neural models that allow detecting and quantifying the state of the driver’s eyes, mouth, and head posture. From these values, a fusion technique is presented that allows inferring the state of the driver’s head while driving. The effectiveness of this proposal is quantified using common metrics from the literature, which suggest a performance greater than \(91\%\) .