Cash Eye – Currency Detection For Blind People Using Machine Learning | IJCT Volume 13 – Issue 2 | IJCT-V13I2P74

In an increasingly digital and accessibility-driven world, visually impaired individuals continue to face significant challenges in conducting everyday financial transactions due to difficulty in identifying the denominations of Indian currency notes. Although the Reserve Bank of India incorporates tactile markers on banknotes, these features often deteriorate over time, reducing their effectiveness and forcing users to rely on external assistance. Existing mobile applications offer partial solutions but typically depend on cloud-based inference, creating barriers for users in areas with limited or inconsistent internet connectivity. The Cash Eye project presents a fully offline, real-time Indian currency recognition system designed to empower visually impaired individuals with greater financial independence. The system integrates deep learning, computer vision, and offline text-to-speech technology to provide instant denomination detection using only a standard camera. A Convolutional Neural Network (CNN) trained on diverse currency images forms the core of the system, while OpenCV enables efficient image acquisition and preprocessing. The TensorFlow-based model processes the input locally, and pyttsx3 generates immediate audio feedback without requiring any network connection. This research emphasizes accessibility, affordability, and practical usability by ensuring that the system operates smoothly on low-cost hardware and resource-constrained environments. The results demonstrate that the proposed approach delivers high accuracy, low latency, and reliable performance, offering a scalable assistive solution that supports financial autonomy for visually impaired users.