Blind person assistance

This project integrates traffic sign detection, object detection, real-time video processing, and fall detection. It uses a combination of OpenCV, machine learning, Firebase Realtime Database, and PyQt5 for GUI development.

---

Features

1. Traffic Sign Detection: Recognizes road signs from a video stream using a trained deep learning model.
2. Object Detection: Detects common objects in the video feed using a COCO pre-trained model.
3. Fall Detection Alert: Monitors a Firebase Realtime Database for fall detection events and sends WhatsApp alerts with the location.
4. Voice Announcements: Uses text-to-speech to announce detected objects and road signs.
5. GUI Interface: Provides a PyQt5-based graphical interface to display the live video feed and detection results.

---

Installation

1. Clone the repository:

   git clone https://github.com/your-username/traffic-sign-object-detection.git
   cd traffic-sign-object-detection

2. Install the required dependencies:

   pip install -r requirements.txt

3. Add your Firebase credentials file:

   • Place your Firebase service account JSON file in the project directory and name it credentials.json.

4. Ensure the following files are present:

   • model_trained.p: Pre-trained traffic sign detection model.
   • coco.names: Class labels for COCO object detection.
   • ssd_mobilenet_v3_large_coco_2020_01_14.pbtxt: Configuration file for the object detection model.
   • frozen_inference_graph.pb: Weights for the object detection model.

---

How to Run

1. Start the program:

   python main.py

2. The GUI will display the live video feed, detected traffic signs, and objects.

---

Usage Guide

Key Features

• Press the "Trigger Function" button in the GUI to save the current video frame and run object detection.
• To stop the application, press q in the video window.

WhatsApp Alerts

The application sends a WhatsApp message using pywhatkit when a fall detection event is detected in Firebase. Ensure you have WhatsApp Web linked to your phone.

---

Directory Structure

traffic-sign-object-detection/
├── credentials.json          # Firebase service account credentials
├── model_trained.p           # Pre-trained traffic sign model
├── coco.names                # Class labels for COCO
├── ssd_mobilenet_v3_large_coco_2020_01_14.pbtxt  # Object detection config
├── frozen_inference_graph.pb # Object detection weights
├── requirements.txt          # Python dependencies
├── main.py                   # Main application script
└── Project_Folder/           # Folder to save captured frames

---

Known Issues and Limitations

• Only .wav files are supported.
• Large audio files may take longer to process.
• Ensure the output directory is not empty; otherwise, the process will stop.

---

Customization

Updating the Detection Model

• Replace model_trained.p with a new traffic sign model if needed.
• Update the coco.names file to include additional object labels.

Modifying Firebase Integration

• Change the Firebase database reference in the code to suit your database structure:

  ref = db.reference("/fallStatus/fall_detected")

WhatsApp Notification

• Update the phone number in the script to your desired recipient:

  phone_number = "+919880738671"  # Replace with the recipient's number

---

Dependencies

• Python 3.6+
• OpenCV
• Numpy
• PyQt5
• Pyttsx3
• PyWhatKit
• PyAutoGUI
• Firebase Admin SDK

---

Contributing

Contributions are welcome! Please open an issue or submit a pull request.

---

License

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