The Real-Time Driver Monitoring System (RTDMS) is a project aimed at improving road safety by employing Artificial Intelligence, Machine Learning, and Computer Vision technologies. The system focuses on detecting driver fatigue and distractions in real time by analyzing visual and behavioral cues such as eye movement, head posture, facial expressions, and object detection.
- Real-Time Monitoring: Detects driver fatigue and distractions using live video feeds.
- Audio-Visual Alerts: Provides immediate feedback to the driver to prevent accidents.
- Environment Adaptability: Robustness in well-lit and medium-lit lighting conditions.
- Mobile-Friendly: Can be used via your smartphone.
- Operating System:
- Compatible with Linux, Windows.
- Ubuntu 20.04+ is recommended for stability with Python, OpenCV, and TensorFlow.
- Python Version:
- Python 3.9.10 (used in development) or any Python 3.6+ version that supports the required libraries.
- Libraries and Frameworks:
- Install the following libraries via
pip:pip install opencv-python dlib numpy mediapipe tensorflow flask
- Ensure that
shape_predictor_68_face_landmarks.dat(required by dlib) is present in the working directory.
- Install the following libraries via
-
Camera:
- A webcam (built-in or external) capable of delivering at least 720p resolution at 30 FPS.
- Higher resolution (1080p) is preferable for better accuracy in facial landmark and object detection.
-
Processor:
- Minimum: Intel i5 (6th generation or equivalent AMD processor).
- Recommended: Intel i7 or better, or an equivalent AMD Ryzen processor for smooth real-time processing.
-
RAM:
- Minimum: 4GB (might work but may face delays).
- Recommended: 8GB or more for handling simultaneous processes (e.g., running Flask, TensorFlow Lite, and OpenCV).
-
GPU (Optional but Beneficial):
- If available, a dedicated GPU (like NVIDIA GeForce GTX 1050 or better) will accelerate TensorFlow Lite inference and OpenCV processing.
- For environments without GPUs, the system can run on the CPU but may experience slightly higher latency.
- First ensure you meet the Hardware and software requirements mentioned above
- Download the folder Local Code
- Run the file "app.py"
- Click the "Enable Audio" Button at the bottom of the screen to start detection
- Open the site : RTDMS
- Allow camera access permission
- Click the "Enable Audio" Button below the dashboard to start detection
- Increase your phone volume
- A stable network connection is recommended for better stability
-
The application alerts users based on the following observations :
- Driver Drowsy : Eyes closed or squinting
- Head Up : Head tilted up for more than 2 seconds
- Head Down : Head tilted down for more than 2 seconds
- Yawning : Mouth open 2 seconds
- Phone In Use : Phone detected in frame
-
Images :
-
- This project, Real-Time Driver Monitoring System, was developed as part of the Mini Project Work course (Course Code: 24AM5PWMPW) during the 5th semester of the Artificial Intelligence and Machine Learning program at BMS College of Engineering.
- Team Members:
- Pranav Veeraghanta (1BM22AI092)
- Shreyas Sachin Kshatriya (1BM22AI125)
- Mitesh J Upadhya (1BM22AI075)
- Shrujal Srinath (1BM22AI127)
For details about the system's architecture, literature review, and methodologies, kindly refer to the Project Report.
If you would like to contribute, report issues, or have any questions about the project, feel free to reach out to us at pranavv.ai22@bmsce.ac.in, shreyassk.ai22@bmsce.ac.in, mitesh.ai22@bmsce.ac.in. Contributions are always welcome!
This software was developed to improve road safety and serve the greater good. Users are encouraged to focus on non-profit and community-driven applications of this technology.
This project is licensed under the MIT License:
- MIT License: A permissive license allowing free use, modification, and distribution for both personal and commercial purposes, with attribution.
For more details, see the MIT License file included in this repository.