Obtaining tire tread model from its real world photo

Tire prints are among the most common types of evidence at crime scenes. Therefore they represent important evidence, that can be essential for a culprit reveal. The important characteristics of a tire tread is its wearing down. This process may affect basic shape of tire tread print.In the paper we present experiment, in which we use convolutional neural network to get 3D model of tire tread from its 2D photography. The model may be later used to simulate wear down of tire tread and thus to estimate new shape of its print.