Alert button

Synthesising Dynamic Textures using Convolutional Neural Networks

Add code
Bookmark button
Alert button
Feb 22, 2017
Christina M. Funke, Leon A. Gatys, Alexander S. Ecker, Matthias Bethge

Figure 1 for Synthesising Dynamic Textures using Convolutional Neural Networks
Figure 2 for Synthesising Dynamic Textures using Convolutional Neural Networks
Figure 3 for Synthesising Dynamic Textures using Convolutional Neural Networks
Figure 4 for Synthesising Dynamic Textures using Convolutional Neural Networks

Share this with someone who'll enjoy it:

Twitter IconFacebook IconLinkedin IconWhatsapp IconMessenger Icon

Here we present a parametric model for dynamic textures. The model is based on spatiotemporal summary statistics computed from the feature representations of a Convolutional Neural Network (CNN) trained on object recognition. We demonstrate how the model can be used to synthesise new samples of dynamic textures and to predict motion in simple movies.

View paper onarxiv icon

Share this with someone who'll enjoy it:

Twitter IconFacebook IconLinkedin IconWhatsapp IconMessenger Icon