Money Detection System

This repository contains a computer vision-based money detection system designed to assist visually impaired individuals by identifying the value of bills shown to a camera. The system uses YOLOv5 for object detection and includes a text-to-speech feature for vocal feedback. This work was done as an end of the year project at Enstab

Features

• Object Detection: Detects the value of bills using a trained YOLOv5 model.
• Text-to-Speech: Announces the detected bill value aloud to assist visually impaired users.
• Classification Model: Includes a ResNet-50 classification model for additional purposes.

Data Collection

The data used for training was manually captured. Each image of a bill was taken in various lighting conditions and angles to ensure robustness in real-world scenarios.

Data Labeling

Labeling was performed using Label Studio, an open-source data labeling tool. The annotations are saved in .txt format compatible with YOLOv5.

Model Training

To train the YOLOv5 model:

1. Clone the YOLOv5 repository and navigate to it:

   git clone https://github.com/ultralytics/yolov5
   cd yolov5
   pip install -r requirements.txt

2. Prepare the data folder structure and split the data into training and validation sets as shown in the notebook.
3. Run the training command:

   python3 train.py --img 243 --epochs 30 --data dataset.yaml --weights yolov5s.pt --cache

4. Note: Make sure to apply necessary modifications to the YOLOv5 repository to suit the specific requirements of this project.

Detection

The detection script is implemented in detect.py. It uses the trained YOLOv5 model to identify bills in real-time. The script also includes a text-to-speech feature, which announces the detected value to the user.

To run detection:

python detect.py --source 0  # Replace 0 with the path to the video file or camera index

Modifications to YOLOv5 Repository

Several changes were made to the YOLOv5 codebase to integrate the text-to-speech feature and optimize the detection process. Please review the detect.py file for specific changes.

Additional Models

The repository also includes a simple classification model based on ResNet-50, available in the model_classification notebook.

Requirements

• Python 3.x
• PyTorch
• YOLOv5 dependencies
• Text-to-Speech libraries (e.g., pyttsx3)

Future Work

• Enhance model accuracy with more diverse data.
• Expand support to detect additional currencies.

Contributing

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

Contact

For any inquiries, please contact Rayen Souli:

• Email: [email protected]
• LinkedIn: Rayen Souli

License

This project is licensed under the ENSTAB License.