BlindScan

BlindScan: Vision for the visionless

Globally, the World Health Organization estimates that approximately 285 million people are visually impaired, of which 39 million are blind. In India alone, the visually impaired population exceeds 15 million, making it one of the countries with a significant number of individuals facing visual challenges. These people often encounter obstacles in managing financial transactions independently, primarily due to the inability to distinguish between different denominations of currency notes.

The visually impaired frequently rely on others to identify currency notes for them, posing risks of financial exploitation and errors. Moreover, the tactile features of Indian rupees can be subtle and difficult to identify, especially for those not accustomed to them or when the notes are worn.

This project proposes a new mobile application that works without any login page or without and particular difficulty or data collection. This will be a simple app which harnesses the power of Convolutional Neural Networks (CNNs) to identify and distinguish between objects. Proposed solution will be in an android application format which will be easy for the user to use.

Important Instructions for BlindScan Project

1. Final Trained Model:

   • The final trained model is saved in the ResNet.ipynb notebook.

2. Additional Models:

   • Besides the ResNet model, the project also utilizes other models, including Inception-v4.

3. Running the Computer Vision Code:

   • To execute the computer vision code, use the main.py script.

4. Setting Up the Environment:

   • Ensure you switch to a working virtual environment.
   • Reinstall all necessary dependencies. Specifically, make sure the following packages are installed:
     • OpenCV
     • TensorFlow