Please use this identifier to cite or link to this item: http://hdl.handle.net/2080/4369
Title: D-Alarm: An Efficient Driver Drowsiness Detection and Alarming System
Authors: Borah, Biju
Mukherjee, Shyamapada
Keywords: Artificial Intelligence
Deep learning
Edge Computing
Internet of Things
Issue Date: Dec-2023
Citation: 6th International Conference on Advances in Science and Technology (ICAST), Mumbai, Indian 8-9 December 2023
Abstract: In the coming years, the demand for IoT devices connecting our homes, cities, and industries will grow significantly. We now inhabit a world where interconnected smart devices can be controlled through single commands, fostering an era of automation, particularly in smart homes. IoT is at the forefront of this transformation, with many of these devices relying on machine learning models to decode data and make accurate predictions. However, as the number of connected devices escalates, network congestion becomes a challenge. To address this, incorporating machine learning intelligence into edge devices is essential. Edge computing, which involves processing data closer to its source, reduces latency and enables faster analysis. Consequently, IoT data can be gathered and processed at the edge, reducing the reliance on cloud-based solutions. Yet, deploying machine learning models on small IoT microcontroller units presents memory and computation constraints. This paper introduces a deep learning model for driver engagement recognition in real-time video streams, alongside optimization techniques for deployment on edge devices like Nvidia Jetson. The goal is to facilitate the execution of machine learning models on compact, low-power integrated circuits, addressing this critical bottleneck
Description: Copyright belongs to proceeding publisher
URI: http://hdl.handle.net/2080/4369
Appears in Collections:Conference Papers

Files in This Item:
File Description SizeFormat 
2023_ICAST_SMukherjee_D-Alarm.pdf3.94 MBAdobe PDFView/Open    Request a copy


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.