IMT School for Advanced Studies Lucca

Increasing Accessibility of Online Board Games to Visually Impaired People via Machine Learning and Textual/Audio Feedback: The Case of "Quantik"

This research addresses the accessibility barriers faced by visually impaired individuals when playing online board games, which are frequently designed exclusively for sighted users.

• A machine learning-based assistive interface was developed to enable visually impaired users to play the abstract strategy game Quantik by providing real-time audio and textual feedback.
• The system utilizes a convolutional neural network trained on 500 images per class to classify game pieces by shape and color, achieving a 94.5% validation accuracy.
• The interface uses cursor or finger tracking to identify game pieces and triggers distinct audio cues, such as gongs or train sounds with varying pitches, to help players distinguish between their pieces and those of an opponent.