Skip to content

README.md

Latest commit

 

History

History
50 lines (42 loc) · 2.41 KB

README.md

File metadata and controls

50 lines (42 loc) · 2.41 KB

TestModel – Driver Monitoring System

    This repository contains the implementation of DMS‑HybridNet, a hybrid deep learning architecture designed for real‑time driver monitoring under challenging conditions such as low‑light environments and occlusion. The system leverages CNNs, BiLSTMs, and Attention mechanisms to efficiently detect driver states, with deployment support through a TensorFlow Lite for mobile devices.

⚠️Note on Model Usage

This repository currently includes an experimental implementation of DMS‑HybridNet (CNN + BiLSTM + Attention). Please be aware that:

  • This is not the final approach we will use in our project.
  • The model here is only a test run to check feasibility with my own dataset.
  • The scripts and dataset are meant for practice and validation, not for production deployment.

Model Architecture

DMS‑HybridNet combines three key components:

  • CNN (Convolutional Neural Networks) → Extracts spatial features from driver facial images.
  • BiLSTM (Bidirectional Long Short‑Term Memory) → Captures temporal dependencies across video frames.
  • Attention Mechanism → Focuses on the most relevant features (e.g., drowsiness cues).

Repository Structure

Data Handling

  • collect_data.py → Collects raw driver data (camera/sensor input).
  • clean_data.py → Cleans and preprocesses datasets.
  • extract_features.py → Extracts features for training.

Testing & Deployment

  • test_camera.py → Tests camera input.
  • test_setup.py → Verifies environment setup.
  • test_tflite.py → Runs TensorFlow Lite model tests.
  • live_test.py → Real‑time driver monitoring demo.

Datasets

  • driver_dataset.csv → Raw dataset.
  • driver_dataset_clean.csv → Cleaned dataset.

Usage Flow

  • Data Collection → Run collect_data.py
  • Data Cleaning → Run clean_data.py
  • Feature Extraction → Run extract_features.py
  • Model Training → Train DMS‑HybridNet with extracted features
  • Testing & Deployment → Use test_tflite.py and live_test.py for real‑time monitoring

Installation

  • git clone https://github.com/JulianaMancera/TestModel.git
  • cd TestModel
  • python -m venv venv
  • source venv/bin/activate # On Linux/Mac
  • venv\Scripts\activate # On Windows
  • pip install -r requirements.txt

Applications

  • Real‑time driver drowsiness detection
  • Monitoring under low‑light conditions