Volumental + Postdoc Erik Bylow develop AI for at-home mobile scanning and recommendations

Important results from the project

** Denna text är maskinöversatt ** The project was implemented in 2020 as part of a larger project where Volumental is developing a mobile app for shoe consumers to collect video around their feet in a home environment. The purpose of the larger project is to be able to recommend shoes with Volumental´s existing proprietary system by obtaining the dimensions of the end user´s feet from neural networks. The goal of this project was to improve the system´s neural network to achieve the desired accuracy by 2020.

Expected long term effects

** Denna text är maskinöversatt ** The result of this project was that Volumental built software that can analyze the effect of various disturbances on the above-mentioned neural networks by creating synthetic input data and visualizing the effect of this in output data. This synthetic data helped to improve the system´s neural network by allowing the team to learn from the various sources of error that existed in the system and by producing more data that improved the system´s neural network.

Approach and implementation

** Denna text är maskinöversatt ** Work took place in an agile way of working that was continuously quality checked. The project followed the original idea and plan to a large extent but could also be adapted when the team learned more about different sources of error. The system developed in the project is still in use and we are now in 2021 close to reaching the desired accuracy and the project and its results have contributed to this.