Distracted Driver Detection is a project aimed at enhancing road safety by using machine learning and computer vision techniques to detect distracted driving behaviors. This repository contains the code and documentation for the project.
Distracted driving is a significant concern as it can lead to accidents, injuries, and even fatalities. This project focuses on developing a solution to detect distracted drivers in real-time using advanced technologies like computer vision and machine learning.
The project uses a dataset from the State Farm Distracted Driver Detection Challenge on Kaggle. The dataset includes around 22,000 images, with a training set of around 17,000 images and a test set of around 5,000 images. Each image is labeled with one of 10 different types of distractions, such as texting, talking on the phone, or reaching behind.
The project's methodology involves data collection, data pre-processing, and model training. Data augmentation techniques are used to increase the diversity of the training data. The models include a custom CNN, VGG19, and EfficientNetB3. Transfer learning is applied using pre-trained models.
Model | Accuracy (Epoch 20) | Validation Accuracy (Epoch 20) |
---|---|---|
Custom CNN | 0.8475 | 0.9025 |
VGG19 | 0.5659 | 0.5579 |
EfficientNetB3 | 0.9170 | 0.9462 |
The project's results show that transfer learning with EfficientNetB3 achieved high accuracy in detecting distracted drivers. The custom CNN model also performed well, demonstrating the effectiveness of the project's architecture and data preparation techniques.
PS: there is Something wrong with the VGG19 model, This could be due to issues related to the model's architecture, hyperparameters, or data preprocessing. It's important to thoroughly investigate these aspects to enhance the model's performance(will be Updated soon).
The Distracted Driver Detection project holds significant promise for improving road safety. The use of machine learning and transfer learning approaches allows for real-time detection of distracted driving behaviors. This project could lead to the development of driver distraction alert mechanisms that can prevent collisions and save lives on the road.
Future development could include integrating the distracted driver detection system with other driving assistance systems and further increasing the accuracy of the detection system.