- mission carried out within Bandai Namco Studios
- create an interface allowing to draw on it and to save the drawing and its label as new data (a man, a car, a house...)
- Process and save the drawings to optimize the learning phase.
- code an ANN that will predict what is the drawing I made thanks to my previous drawings.
Finally, to add some precision, I added some complexity in the pixel’s reduction and introduced an "average pixel" so that the neural network has some more precise information while learning. Indeed, instead of having black and white pixels, I now have different shades of grey pixels, allowing the ANN to have more precision and more data on each drawing.
