DiTo97 / stunning-potato

Speech Processing and Recognition (101803) final project at UniGe

Stunning Potato

Stunning Potato is intended to be part of a bigger project on the Android platform which aims at jointly tackling Computational vision (90539) and Speech processing (101803) final projects.

This project focuses on the development of tools (i.e., Android apps and AI models) to improve the social engagement of visually-impaired people with the external world.

Please note: Due to time constraints we did not manage to plug the AI models into the app:

• Dense image captioning: https://github.com/DiTo97/kwl-dense-captioning

Folder structure

• apk folder contains the downloadable APK file StunningPotato.apk (min API 26)
• report folder contains the written PDF report

Why "Stunning Potato"?

We were looking for a name that could convey true emotions to the user by means of thorough descriptions of the environment surrounding him at any given moment. It doesn’t really matter what a visually-impaired person may be looking at, neither noticeable or common. What does truly matter instead, is how vivid the idea that we can give him of the outer world is, to whom even a simple potato might be the most stunning thing ever.