https://github.com/titu1994/keras-non-local-nets

Keras implementation of Non-local Neural Networks
https://github.com/titu1994/keras-non-local-nets

convolutional-neural-networks keras neural-network non-local

Last synced: 6 days ago
JSON representation

Keras implementation of Non-local Neural Networks

• Host: GitHub
• URL: https://github.com/titu1994/keras-non-local-nets
• Owner: titu1994
• License: mit
• Created: 2017-11-23T05:04:13.000Z (about 7 years ago)
• Default Branch: master
• Last Pushed: 2024-05-13T05:00:38.000Z (9 months ago)
• Last Synced: 2025-01-19T21:11:43.892Z (14 days ago)
• Topics: convolutional-neural-networks, keras, neural-network, non-local
• Language: Python
• Size: 267 KB
• Stars: 290
• Watchers: 17
• Forks: 100
• Open Issues: 8
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Keras Non-Local Neural Networks

Keras implementation of Non-local blocks from [[1]](https://arxiv.org/abs/1711.07971).

- Support for `"Gaussian"`, `"Embedded Gaussian"` and `"Dot"` instantiations of the Non-Local block. 

- Support for variable shielded computation mode (reduces computation by N**2 x, where N is default to 2)

- Support for `"Concatenation"` instantiation will be supported when authors release their code.

# Usage Templates

The script `non_local.py` contains the `NonLocalBlock` instance which takes in an input tensor and wraps a non-local block around it.

```python

from non_local import NonLocalBlock

from tensorflow.keras.layers import Input, Conv1D, Conv2D, Conv3D

ip = Input(shape=(...))  # input tensor with an "N" rank order of 3, 4 or 5

x = ConvND(...)         # convolution operation with aforementioned rank 

...

non_local_block = NonLocalBlock(intermediate_dim=None, compression=2, mode='embedded', add_residual=True)

x = non_local_block(x)

...

```

The script `non_local_layerstyle.py` contains the `NonLocalBlock` **layer** which takes in an input tensor and wraps a non-local block around it. Made to facilitate the neural network builder using the Sequential method. 

```python

from non_local_layerstyle import NonLocalBlock

from tensorflow.keras.layers import Input, Conv1D, Conv2D, Conv3D

# Define the input shape

input_shape = (...)  # shape of input tensor

model = Sequential()

model.add(ConvND(...))         # convolution operation with an "N" rank order of 3, 4 or 5

...

model.add(NonLocalBlock(intermediate_dim=None, compression=2, mode='embedded', add_residual=True))

...

```

# Basic block

From [[1]](https://arxiv.org/abs/1711.07971), a basic Non-Local block with the Embedded Gaussian instantiation has the below logic:



1. Xiaolong Wang, Ross Girshick, Abhinav Gupta, Kaiming He. "Non-local Neural Networks." arXiv:1711.07971 [cs.CV], 21 Nov 2017. [Link](https://arxiv.org/abs/1711.07971)