Program Type
Undergraduate
Faculty Advisor
Kaiman Zeng
Document Type
Presentation
Location
Face-to-face
Start Date
25-4-2023 9:40 AM
Abstract
In 2020, distracted driving claimed 3,142 lives in the United States[1]. Using current technology, our project aims to prototype a device that uses computer vision to crack down on this dangerous behavior while working with insurance agencies that provide incentives for customers to adopt the device. At the heart of this project is a deep learning algorithm trained to classify images on a set of 10 classes: safe driving, texting - right, talking on the phone - right, texting - left, talking on the phone - left, operating the radio, drinking, reaching behind, hair and makeup, and talking to passenger. This algorithm is deployed onto a microcontroller, which is mounted above the passenger side window with a connected camera and power supply, taking images like the one shown above. In short, the algorithm classifies periodically taken images and determines the motorist’s safe-driving performance. The driver's performance is then sent to the insurance company through a mobile application. From there, the insurance company will adjust the user’s insurance rate based on their safe driving performance. Thus, an incentive for drivers is made to adopt the product and practice safe driving more thoroughly.
Recommended Citation
Caja, Tristan; Riney, Lizzi; Gregurek, Chance; and Heikes, Wesley M., "Distracted Driving Detection System" (2023). ATU Research Symposium. 41.
https://orc.library.atu.edu/atu_rs/2023/2023/41
Included in
Distracted Driving Detection System
Face-to-face
In 2020, distracted driving claimed 3,142 lives in the United States[1]. Using current technology, our project aims to prototype a device that uses computer vision to crack down on this dangerous behavior while working with insurance agencies that provide incentives for customers to adopt the device. At the heart of this project is a deep learning algorithm trained to classify images on a set of 10 classes: safe driving, texting - right, talking on the phone - right, texting - left, talking on the phone - left, operating the radio, drinking, reaching behind, hair and makeup, and talking to passenger. This algorithm is deployed onto a microcontroller, which is mounted above the passenger side window with a connected camera and power supply, taking images like the one shown above. In short, the algorithm classifies periodically taken images and determines the motorist’s safe-driving performance. The driver's performance is then sent to the insurance company through a mobile application. From there, the insurance company will adjust the user’s insurance rate based on their safe driving performance. Thus, an incentive for drivers is made to adopt the product and practice safe driving more thoroughly.