DFFNet: An IoT-perceptive Dual Feature Fusion Network for General Real-time Semantic Segmentation.
Xiangyan Tang, Wenxuan Tu, Keqiu Li, Jieren Cheng.
Information Sciences, 565: 326-343, 2021.
All rights reserved. Licensed under the Apache License 2.0
The code is released for academic research use only. For commercial use, please contact [wenxuantu@163.com].
Clone this repo.
https://github.com/WxTu/DFFNet.git- Windows or Linux
- Python3
- Pytorch(0.3+)
- Numpy
- Torchvision
- Matplotlib
We use Cityscapes, Camvid and Helen datasets. To train a model on these datasets, download datasets from official websites.
Our backbone network is pre-trained on the ImageNet dataset provided by F. Li et al. You can download publically available pre-trained MobileNet v2 from this website.
data/Dataset.py: processes the dataset before passing to the network.model/DFFNet.py: defines the architecture of the whole model.model/Backbone.py: defines the encoder.model/Layers.py: defines the MFFM, LSPM, and others.utils/Config.py: defines some hyper-parameters.utils/Process.py: defines the process of data pretreatment.utils/Utils.py: defines the loss, optimization, metrics, and others.utils/Visualization.py: defines the data visualization.Train.py: the entry point for training and validation.Test.py: the entry point for testing.
Any discussions or concerns are welcomed!
If you use this code for your research, please cite our papers.
@article{Tang2021DFFNet,
title={DFFNet: An IoT-perceptive Dual Feature Fusion Network for General Real-time Semantic Segmentation},
author={Xiangyan Tang and Wenxuan Tu and Keqiu Li and Jieren Cheng},
journal={Information Sciences},
volume={565},
pages={326-343},
year={2021}
}
https://github.com/ansleliu/LightNet
https://github.com/meetshah1995/pytorch-semseg
https://github.com/zijundeng/pytorch-semantic-segmentation
https://github.com/Tramac/awesome-semantic-segmentation-pytorch