Drivers’ behaviours detection with privacy protection using deep learning techniques

• Main Author: Xing, Tong
• Other Authors: Tay, Wee Peng
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2021
• Subjects: Engineering::Electrical and electronic engineering::Computer hardware, software and systems
• Online Access: https://hdl.handle.net/10356/149893
• Institution: Nanyang Technological University

The aim of this project is to detect driver’s anomaly behaviour with privacy protection. In normal life, the driver may be distracted from other behaviours and may not focus on driving, which causes serious accidents. According to the research conducted by CDC safety division, distracted driving has become the top 1 cause of road accidents in the United States. Over 40 thousand people get injured and 6,000 people lose their lives because of distracted driving every year. The distracted behaviours may include drinking water, having a phone call, texting, reaching behind etc. Therefore, it’s critical to detect the drivers’ anomaly behaviours if possible so that the number of traffic accidents can be further eliminated. This can be done through classification from deep learning techniques with trained model by using Convolutional Neural Network (CNN). However, when the pictures of drivers are collected, there are possibilities that the drivers’ privacy information, such as gender, age, ethnicity may be used and leaked. In fact, privacy leakage has become a serious concern in these years and people begin to pay more attention to data privacy. Therefore, this project aims to propose a privacy preserving framework that serves as a method to sanitize the data where the privacy information, such as age and gender will be removed. After the process of sanitizing, the sanitized data will be further classified and various anomaly behaviours of drivers will be detected.