BrailleCart: AI-Powered Grocery Assistance for the Visually Impaired

Introduction

BrailleCart is an AI-powered system designed to assist visually impaired individuals in identifying and learning about grocery items in real-time. Combining advanced AI technologies like YOLOv8n for object detection, OCR for extracting product details, and LLaMA-based conversational text-to-speech, BrailleCart enhances the shopping experience through a simple and accessible interface.

Dataset

Dataset download link

Key Features

• Real-Time Object Detection: Utilizes YOLOv8n for precise and quick identification of grocery items.
• Text Recognition (EasyOCR): Extracts product names, prices, and other details from labels.
• Conversational Feedback: Uses LLaMA to generate natural language descriptions of detected items.
• Audio Accessibility: Provides real-time audio feedback with Google Text-to-Speech (gTTS).
• User-Friendly Interface: Streamlit-based UI for seamless interaction.

Steps to Run the Appliction

• Clone the repository:

  git clone https://github.com/sudip0789/BrailleCart.git
  cd BrailleCart
  

• Replace the hugging face token with your own HF token

• Start the application:

  streamlit run braillecart_app.py
  

The detailed report and results are available here