Distracted driving is a significant cause of accidents, resulting in thousands of deaths and injuries each year. This project aims to leverage computer vision and machine learning to detect inattentive drivers using dashboard cameras, ultimately reducing accidents caused by distracted driving. The State Farm Distracted Driving image dataset, comprising over 100,000 labeled images, forms the basis of our research. We explore various state-of-the-art machine learning algorithms to classify images into ten different actions, such as safe driving, texting, talking on the phone, and more. The project consists of key phases including data exploration, model building, website development, and model deployment. In addition to the models, we built a website to showcase our work. The website allows users to upload an image, which is then processed by our trained model to predict the driver's behavior.
- Python
- TensorFlow
- Keras
- OpenCV
- Flask (for website deployment)
- HTML/CSS (for website development)
- Streamlit (for deploying the model)
- Data Exploration & Visualization
- Convolutional Neural Networks (CNNs)
- Transfer Learning
- Model Building & Optimization
- Web Development
- Deployment & Hosting