Road accidents have increased in recent years for a variety of causes. Distracted driving is a leading cause of car crashes, and unfortunately, many of these crashes end in fatalities. In this study, we are using computer vision with machine learning, to provide a unique technique to determine whether a car driver is driving the car safely or not. In this study, two feature extraction approaches SIFT and ORB were implemented, with an emphasis on the driver's location, expression, and behavior. These two techniques effectively extract the features from the images in the dataset. This was followed by PCA to reduce dimensions. The implementation work further deployed several models including Decision Tree, Random Forest, KNN, and SVM to identify distracted driving behaviors. When using the ORB feature extractor, more accuracy was achieved in comparison with SIFT. The highest accuracy achieved was 90.75%, using KNN classifier. In the case of SIFT, the SVM RBF classifier gave the highest accuracy of 78.00%. It appears from the experiments that the suggested system has the ability to aid drivers in practicing safe driving behaviors.
Vision-Based Distracted Driver Detection Using a Fusion of SIFT and ORB Feature Extraction
Lect. Notes Electrical Eng.
International Conference on Security, Privacy and Data Analytics ; 2022 ; Surat, India December 13, 2022 - December 15, 2022
2023-08-19
16 pages
Article/Chapter (Book)
Electronic Resource
English