Implementing a sketch prediction app

Abstract

[en] The main topic of this work is object classification applied to the sketch domain. Starting from a huge set of object drawings, we have studied the problem of recognizing them in order to build a drawing app. This project, then consists in three main blocks, multiplatform app development that let users draw, deep neural network in order to recognize the drawings and an api that recieves the image and outputs the prediction. The true challenge of this project is then to ensure a relaible recognition alltogether.