Phase 1 - Autonomy Perception

A real-time Advanced Driver Assistance System (ADAS) perception stack built with Python. Features object detection, lane tracking, and driver monitoring, optimized for unstructured traffic environments (Indian Roads).

Quick Start

# Clone repository
git clone https://github.com/Sherin-SEF-AI/Phase1-Autonomy-Perception.git
cd Phase1-Autonomy-Perception

# Create virtual environment
python3 -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Run application with all features
python3 run_all_features.py

Features

Core Detection & Tracking

• Object Detection - YOLOv8-based detection (people, vehicles, bikes)
• Multi-Object Tracking - Kalman filter-based tracking with unique IDs
• Lane Detection - Real-time lane line detection
• Collision Warning - Forward collision warnings
• Safe Distance Monitoring - Tailgating detection

Traffic & Road Analysis

• Traffic Sign Recognition - Stop, yield, speed limits
• Traffic Light Detection - Red/yellow/green state detection
• Road Debris Detection - Obstacle and debris detection
• Pothole Detection - Road surface damage detection
• Parking Space Detection - Available/occupied space marking

AI-Powered Features

• Scene Classification - Road/highway/urban/parking detection
• Vehicle Type Classification - Car/truck/bus/motorcycle
• Weather Detection - Clear/rainy/foggy/snowy conditions
• Driver Attention Monitoring - Drowsiness & distraction detection
• Emergency Vehicle Detection - Police/ambulance/fire truck
• Night Vision Enhancement - Low-light image enhancement

Advanced Analytics

• Optical Flow Analysis - Dense motion field visualization
• Motion Prediction - Future trajectory prediction
• Driving Behavior Analysis - Aggressive/normal/cautious scoring
• Pedestrian Pose Estimation - 33-point body keypoints
• License Plate Detection - Vehicle plate recognition
• Object Density Heatmap - Heat map visualization

Visualization & Monitoring

• Statistics Dashboard - Real-time metrics panel
• Performance Graphs - FPS, CPU, object count over time
• Multi-Camera Display - Support for 4 cameras
• Real-time Overlays - Bounding boxes, tracking IDs, predictions

Available Versions

Version	Features	FPS	Best For
run_all_features.py	20+	8-15	All capabilities showcase
adas_complete_ultra.py	Custom	8-30	Flexible configuration
adas_fast.py	6	25-35	Real-time performance
adas-perception.py	20	20-25	Standard production use

Run with All Features

python3 run_all_features.py

Run with Custom Selection

python3 adas_complete_ultra.py
# Feature selection dialog appears - choose your features

Run Speed-Optimized Version

python3 adas_fast.py

Feature Selection Dialog

The adas_complete_ultra.py version includes a pre-launch dialog where you can:

• Choose exactly which features to enable
• See real-time FPS estimates
• Use quick presets (All, Balanced, Speed, None)
• Save/Load configurations as profiles
• View performance impact for each feature

Quick Presets:

• All Features - Enable everything (8-15 FPS)
• Balanced - Good features without heavy ones (18-22 FPS) - Recommended
• Speed Mode - Essential features only (25-30 FPS)
• Disable All - Core ADAS only (30-40 FPS)

Requirements

• Python: 3.8 or higher
• Operating System: Linux, macOS, or Windows
• RAM: 8GB minimum, 16GB recommended
• CPU: Multi-core processor (i5 or better)
• Camera: Webcam or video file

Python Dependencies

opencv-python>=4.8.0
numpy>=1.24.0
ultralytics>=8.0.0
mediapipe>=0.10.0
wxPython>=4.2.0
filterpy>=1.4.5
Pillow>=10.0.0

Install all dependencies:

pip install -r requirements.txt

Documentation

• START_HERE.txt - First-time setup guide
• HOW_TO_RUN_ALL_FEATURES.txt - Complete guide for all features
• FEATURE_SELECTION_QUICKSTART.txt - Feature selection quick reference
• WHICH_VERSION_TO_RUN.txt - Version selection guide
• ALL_RUN_OPTIONS.txt - All available run options

Usage Examples

Example 1: Basic Usage

# Run with balanced preset (recommended for first-time users)
python3 adas_complete_ultra.py
# Click "Balanced" -> "Start Application"

Example 2: Demo/Presentation

# Run speed-optimized for smooth playback
python3 adas_fast.py

Example 3: Full Feature Showcase

# Run with all 100+ features
python3 run_all_features.py

Example 4: Custom Configuration

# Use feature selection dialog
python3 adas_complete_ultra.py
# Manually select desired features
# Save as profile for reuse

Configuration

Camera Settings

Edit camera settings in the application:

• Number of cameras: 1-4
• Resolution: 640x480, 1280x720, 1920x1080
• FPS: 15, 30, 60

Feature Settings

Toggle features at runtime using GUI checkboxes:

• ULTRA FEATURES section (orange labels)
• AI FEATURES section (green labels)

Or configure before launch using feature selection dialog.

Performance Tips

For best performance:

• Use 1 camera instead of 4
• Use 640x480 resolution for 4x speed boost
• Close background applications
• Use "Balanced" or "Speed Mode" presets
• Disable heavy features if FPS is low

Heavy features (consider disabling):

• Driver Attention Monitoring (-6 FPS)
• Pedestrian Pose Estimation (-7 FPS)
• Parking Space Detection (-5 FPS)
• Optical Flow Visualization (-5 FPS)

Project Structure

Phase1-Autonomy-Perception/
├── adas_complete_ultra.py      # Main app with feature selection
├── run_all_features.py          # Quick launcher (all features)
├── adas_fast.py                 # Speed-optimized version
├── adas-perception.py           # Standard version
├── ultra_features.py            # Advanced detection features
├── ultra_ai_features.py         # Modern AI features
├── adas_ultra_advanced.py       # Core ultra features
├── ultra_visualization.py       # Visualization components
├── advanced_modules.py          # Analytics modules
├── requirements.txt             # Python dependencies
├── examples/                    # Example scripts
│   ├── basic_usage.py
│   └── advanced_analytics.py
└── docs/                        # Documentation files
    ├── START_HERE.txt
    ├── HOW_TO_RUN_ALL_FEATURES.txt
    └── ...

Testing

Run the test suite:

python3 test_ultra_features.py

Test individual features:

python3 examples/basic_usage.py
python3 examples/advanced_analytics.py

Troubleshooting

Application is slow

• Use fewer cameras (1 instead of 4)
• Lower resolution (640x480)
• Disable heavy features
• Try adas_fast.py

Features not working

• Check that all dependencies are installed
• Verify camera is connected
• Check console for error messages

Camera not detected

• Check camera permissions
• Try different camera index (0, 1, 2, etc.)
• Test with: python3 -c "import cv2; print(cv2.VideoCapture(0).read())"

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• YOLOv8 by Ultralytics - Object detection
• MediaPipe by Google - Pose estimation
• OpenCV - Computer vision
• wxPython - GUI framework

Author

Sherin Joseph Roy

• Co-Founder & Head of Products, DeepMost AI
• Location: Bangalore, India
• Website: sherinjosephroy.link
• GitHub: github.com/Sherin-SEF-AI
• LinkedIn: linkedin.com/in/sherin-roy-deepmost
• X (Twitter): x.com/SherinSEF
• Mastodon: mastodon.social/@sherinjoesphroy

Project

Project: Phase 1 - Autonomy Perception Organization: DeepMost AI Repository: github.com/Sherin-SEF-AI/Phase1-Autonomy-Perception

---