https://github.com/titu1994/fully-connected-densenets-semantic-segmentation
Fully Connected DenseNet for Image Segmentation (https://arxiv.org/pdf/1611.09326v1.pdf)
https://github.com/titu1994/fully-connected-densenets-semantic-segmentation
Last synced: 28 days ago
JSON representation
Fully Connected DenseNet for Image Segmentation (https://arxiv.org/pdf/1611.09326v1.pdf)
- Host: GitHub
- URL: https://github.com/titu1994/fully-connected-densenets-semantic-segmentation
- Owner: titu1994
- Created: 2016-12-11T12:51:25.000Z (almost 9 years ago)
- Default Branch: master
- Last Pushed: 2017-03-29T14:42:20.000Z (over 8 years ago)
- Last Synced: 2025-03-18T09:02:51.058Z (7 months ago)
- Language: Python
- Size: 31.3 KB
- Stars: 84
- Watchers: 6
- Forks: 41
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Fully Connected DenseNets for Semantic Segmentation
Fully Connected DenseNet for Image Segmentation implementation of the paper [The One Hundred Layers Tiramisu : Fully Convolutional DenseNets for Semantic Segmentation](https://arxiv.org/pdf/1611.09326v1.pdf)
# Differences
- Use of SubPixelConvolution instead of Deconvolution as default method for Upsampling.
# Usage :
Simply import the `densenet_fc.py` script and call the create method:
```
import densenet_fc as dc
model = DenseNetFCN((32, 32, 3), nb_dense_block=5, growth_rate=16,
nb_layers_per_block=4, upsampling_type='upsampling', classes=1)
```
# Requirements
Keras 1.2.2
Theano (master branch) / Tensorflow 1.0+
h5py