Project Info

NeuroDrive

This project did not submit a demo video on Devpost.

Inspiration

Randomly, we encountered an unexpected statistic: 1 in 25 adults has dozed off while driving in the previous month. This made us discover that fatigue-related driving is responsible for more than 100,000 accidents annually in the U.S., with a significant number involving long-haul operators. We were amazed at how frequently fatigue and distraction remain overlooked until it's too late — particularly on lengthy journeys. That’s when we started considering if we could create a system that not only comprehends the road but also focuses on the driver. That concept formed the basis for NeuroGuardian.

What it does

Data Entries An EEG headset was utilized to track the driver’s cognitive condition in real time, recording signals indicative of attention and fatigue. In addition, we installed a dual-camera system: one directed at the driver to monitor behavior, and another aimed at the road to collect traffic and lane information. Detection Workflow We incorporated YOLOv8 for identifying objects and utilized UltraFast Lane Detection to analyze the drivable region. These outputs were merged to categorize objects according to the lane zone where they were detected. At the same time, EEG signals were examined to identify states of distraction or fatigue. Results & Notifications If the system identifies distraction or potential risk in the driver's lane, it activates a voice alert through a text-to-speech system (using Gemini). The alert aims to be simple yet impactful, assisting in redirecting the driver’s attention without causing them to feel overloaded

How we built it

NeuroGuardian is an assistant for driver awareness that integrates brainwave signals with computer vision to minimize distraction and fatigue. An EEG headset tracks the driver's concentration in real time, supplemented by a dual-camera system—one directed at the road and the other observing the driver. It identifies lane boundaries and detects objects with YOLOv8, subsequently classifying those objects according to their location in relation to the driving lane. If it senses a lapse in attention or a possible danger in a crucial area, it provides a brief voice alert via a text-to-speech system to redirect the driver’s focus to the road.

Challenges we ran into

We faced issues with training and fine-tuning models due to limited processing power. Additionally, finding data sets for our EEG model was difficult.

Accomplishments we're proud of

We are proud of combining both EEG technology and CV to protect drivers, which hasn't been done before.

What we learned

We learned how to reduce the noise in data. The data and dataset for all our AI models were noisy, so we've learned the importance and a lot of techniques in it while tackling the data analysis.

What's next

We would like to use a smaller BCI device such as Muse2, but the data quality is low, so we also need to improve the machine learning model. We will also need feedback from the users so that we can find further improvements.

NeuroDrive

UC Berkeley AI Hackathon 2025 Project

A real-time driver monitoring system that integrates computer vision and EEG analysis to detect distraction, drowsiness, and potential hazards for enhanced driving safety.

🏆 Project Vision

We win these - Building the future of intelligent driver assistance systems.

🏗️ Architecture

Microservices Design

┌─────────────────┐    ┌─────────────────┐    ┌─────────────────┐
│  Distraction CV │    │  Dizziness EEG  │    │ Object Detection│
│    Port 8001    │    │    Port 8002    │    │    Port 8003    │
└─────────────────┘    └─────────────────┘    └─────────────────┘
         │                       │                       │
         └───────────────────────┼───────────────────────┘
                                 │
                    ┌─────────────────┐
                    │    Main GUI     │
                    │   (Tkinter)     │
                    └─────────────────┘

Core Services

Distraction CV Service (services/distraction_cv.py)
- Real-time eye tracking and attention detection
- Head pose estimation
- WebSocket streaming on port 8001
Dizziness EEG Service (services/dizziness_eeg.py)
- EEG signal processing for drowsiness detection
- Brain wave analysis (Alpha, Beta, Theta)
- WebSocket streaming on port 8002
Object Detection Service (services/object_detection_cv.py)
- Real-time road hazard detection
- Vehicle and pedestrian identification
- WebSocket streaming on port 8003
Main Integration (main.py)
- Tkinter-based GUI for unified monitoring
- Real-time data aggregation from all services
- Central control and visualization

🚀 Quick Start

Prerequisites

Python 3.8+
pip package manager

Installation

Clone the repository

git clone <repository-url>
cd inner-eye-drive

Install dependencies
```
pip install -r requirements.txt
```

Running the System

Start all services (in separate terminals):

# Terminal 1 - Distraction CV
python3 services/distraction_cv.py

# Terminal 2 - Dizziness EEG  
python3 services/dizziness_eeg.py

# Terminal 3 - Object Detection
python3 services/object_detection_cv.py

Launch the main application:

# Terminal 4 - Main GUI
python3 main.py

Test with multi-client (optional):

# View all service outputs simultaneously
python3 multi_client.py

🔧 API Endpoints

Service Health Checks

Distraction CV: GET http://127.0.0.1:8001/
Dizziness EEG: GET http://127.0.0.1:8002/
Object Detection: GET http://127.0.0.1:8003/

WebSocket Endpoints

Distraction CV: ws://127.0.0.1:8001/ws
Dizziness EEG: ws://127.0.0.1:8002/ws
Object Detection: ws://127.0.0.1:8003/ws

📁 Project Structure

inner-eye-drive/
├── main.py                     # Main GUI application
├── multi_client.py            # Multi-service test client
├── client.py                  # Single service test client
├── requirements.txt           # Python dependencies
├── README.md                  # Project documentation
├── services/
│   ├── distraction_cv.py     # Eye tracking & attention detection
│   ├── dizziness_eeg.py      # EEG drowsiness detection
│   └── object_detection_cv.py # Road hazard detection
└── models/                    # ML models (future implementation)

🛠️ Development Status

✅ Completed

Microservices architecture setup
FastAPI + WebSocket infrastructure
Real-time communication between services
Basic Tkinter GUI framework
Multi-service client testing

🚧 In Progress

Computer vision algorithms implementation
EEG signal processing integration
Machine learning model integration
Advanced GUI with real-time visualizations

📋 Planned Features

Real-time driver alerting system
Data logging and analytics
Mobile app integration
Cloud-based monitoring dashboard

🤝 Contributing

This project was developed during the UC Berkeley AI Hackathon 2025.

📄 License

[Add your license information here]

Built with ❤️ at UC Berkeley AI Hackathon 2025

Analysis

Compare with all teams

View

Metric

Figures cover GitHub contributors during the hackathon window. A co-authored commit counts in full for each author, so per-member totals add up to more than the whole-team figures.

Technology

Found in codeClaimed only

FastAPIIn code
PythonIn code
PyTorchIn code
TensorFlowIn code

4 of 4 appear in the indexed code.

AI coding agents

No AI coding agent signals were found in this repository.

Detected from committed agent config files and commit authorship. Absence of a signal is not proof an agent was unused.

Codebase size

Source size

167 KB

Source files

Counts recognized source files only; vendored directories, binaries and lockfiles are excluded, so this is smaller than the repository on disk.

Project Info

NeuroDrive

Inspiration

What it does

How we built it

Challenges we ran into

Accomplishments we're proud of

What we learned

What's next

NeuroDrive

🏆 Project Vision

🏗️ Architecture

Microservices Design

Core Services

🚀 Quick Start

Prerequisites

Installation

Running the System

🔧 API Endpoints

Service Health Checks

WebSocket Endpoints

📁 Project Structure

🛠️ Development Status

✅ Completed

🚧 In Progress

📋 Planned Features

🤝 Contributing

📄 License

Analysis

Technology

AI coding agents

Codebase size

Structure

Application logic

Data & schema

Languages

Dependencies