Classifies 3D meshes taken from the princeton shape benchmark

This projects consists of 2 parts.

1. A jsonrpc server running inside a blender process allowing remote calls for actions like moving the camera and rendering to files.

2. A neural network based on Deepmind's recent work with recurrent attention.[1][2] The network learns how to classify a 3D shape given a few renders of that shape and learns a policy for moving the camera to get the best views for classifying.

[1]Mnih, Volodymyr, Heess, Nicolas, Graves, Alex, and Kavukcuoglu, Koray. Recurrent models of visual attention. arXiv preprint arXiv:1406.6247, 2014

[2] Jimmy Ba, Volodymyr Mnih, Koray Kavukcuoglu. Multiple Object Recognition with Visual Attention. CoRR abs/1412.7755 (2014)

Coming Soon :)