Blind Assist System for Currency Recognition Using Sequential Convolutional Neural Networks

Currency recognition is crucial for assisting visually impaired individuals in identifying and managing money. Visually impaired people are likely to get scammed during their financial transactions. This project helps the visually impaired to recognize and distinguish different currencies. This project employs a deep learning approach, specifically using Convolutional Neural Networks (CNNs), accurately detect and classify currency from images, thereby enhancing accessibility and independence for the blind. The project involves several key steps, beginning with image preprocessing to standardize quality and reduce noise. This includes resizing and normalization to ensure consistency across input data. The CNN architecture incorporates 2D max- pooling and convolutional layers with ReLU activation functions for feature extraction and classification, respectively. The model is trained on a labeled dataset of currency images, optimizing parameters through backpropagation to minimize classification errors. Evaluation on a test dataset measures accuracy, precision, and recall. Once trained, the model can be integrated into real-world applications, providing real-time currency recognition and aiding visually impaired users in day-to-day transactions. The model gave an overall accuracy of 84.03% Continuous updates to the system ensure its effectiveness and usability in diverse environments, contributing to accessibility technologies for the visually impaired..