BankNote‐Net: Open Dataset for Assistive Universal Currency Recognition

Summary

Millions of people around the world have low or no vision. Assistive software applications—including optical character recognition, scene identification, person recognition, and currency recognition—have been developed to help such people engage in day‐to‐day activities. Using supervised contrastive learning, this chapter develops a machine learning model for universal currency recognition. To address the increasing problem of visual impairment, the software community has developed a variety of assistive technologies, such as mobile applications and wearables. BankNote‐Net can serve as a valuable resource for the research community. The database and embeddings generated exhibit robust generalization capabilities. This extends even to currencies that are not part of the BankNote‐Net dataset. Future work should include diversifying the range of data collectors to mitigate existing biases and expand the dataset's utility.