GitHub - D-l-z/3D-Spatial-Transformer-Network_tensorflow: Use the attention mechanism, named ''Spatial Transformer Networks'', to implement 3D reconstruction of 2D single images with Tensorflow.

D-l-z / 3D-Spatial-Transformer-Network_tensorflow Public

Notifications You must be signed in to change notification settings
Fork 0
Star 1

Use the attention mechanism, named ''Spatial Transformer Networks'', to implement 3D reconstruction of 2D single images with Tensorflow.

Apache-2.0 license

1 star 0 forks Branches Tags Activity

Star

Notifications

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 2 Commits
LICENSE		LICENSE
cluttered_3Dmnist.py		cluttered_3Dmnist.py
cluttered_mnist.py		cluttered_mnist.py
readme.py		readme.py
spatial_3Dtransformer.py		spatial_3Dtransformer.py
spatial_transformer.py		spatial_transformer.py
tf_utils.py		tf_utils.py

Repository files navigation

# 3D Spatial Transformer Network

## Introduction
This is the TensorFlow implementation for 2D & 3D Spatial Transformer Network in 
		[1] Jaderberg M, Simonyan K, et al. Spatial transformer networks[C]. International Conference on Neural Information Processing Systems. MIT Press, 2016:2017-2025.

Re-implemented by Lizhen Duan, Jie Li

Reference: [1] https://github.com/daviddao/spatial-transformer-tensorflow
           [2] the code by Xinchen Yan, Arkanath Pathak, Jasmine Hsu, Honglak Lee [I lost the code link, but I know they refer to Orginal implementation in Torch (https://github.com/xcyan/nips16_PTN)] 
		   
#### How to use
run cluttered_3Dminist.py for 3D data
run cluttered_minist.py for 2D data

### Installation
*   TensorFlow
*   matplotlib

### Dataset
2d minist in data/mnist_sequence1_sample_5distortions5x5.npz
3d minist in data/3d-minist/full_dataset_vectors.h5