Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.
gbliao 382da2ee20 | 2 years ago | |
---|---|---|
MMNet.py | 2 years ago | |
README.md | 2 years ago | |
__init__.py | 2 years ago | |
dataset_loader.py | 2 years ago | |
functions.py | 2 years ago | |
fusion.py | 2 years ago | |
res2net50.pth | 2 years ago | |
test.py | 2 years ago | |
train.py | 2 years ago | |
trainer.py | 2 years ago |
MMNet: Multi-Stage and Multi-Scale Fusion Network for RGB-D Salient Object Detection. PDF
Add performance of our MMNet trained on the new COME-Train dataset (8025 images):
https://pan.baidu.com/s/1S2ZT1AGqW0CfwaGFmubbbQ [code: wl4s]
• pytorch 1.3.0+
• torchvision
• PIL
• Numpy
• Download the trained model from here [code: ofcn]
• Download test datasets from here [code: sva4]
• Modify your test_dataroot
and test_datasets
in test.py
• Test the MMNet: python test.py
• Download the train-augment dataset from here [code: haxl]
• Download the pretrained backbone Res2Net(baseWidth = 48, scale = 2) from here
• Modify your train_dataroot
and pre_trained_root
in train.py
• Train the MMNet: python train.py
• Saliency maps mentioned in the paper can be download from here [code: wl4s]
[1] The test_Results are obtained by trained on NJUD & NLPR & DUT (1485+700+800).
[2] The test_results_COME_train are obtained by trained on the new COME-Train dataset (8025).
• The saliency results can be evaluated by using the tool in Matlab
Please cite our paper if you use this repository in your reseach.
@inproceedings{MMNet20,
author = {Liao, Guibiao and Gao, Wei and Jiang, Qiuping and Wang, Ronggang and Li, Ge},
title = {MMNet: Multi-Stage and Multi-Scale Fusion Network for RGB-D Salient Object Detection},
booktitle = {Proceedings of the 28th ACM International Conference on Multimedia},
pages = {2436–2444},
year = {2020}
}
PyTorch implementation of the paper ‘MMNet: Multi-Stage and Multi-Scale Fusion Network for RGB-D Salient Object Detection’ (ACMMM'2020)
Python
Dear OpenI User
Thank you for your continuous support to the Openl Qizhi Community AI Collaboration Platform. In order to protect your usage rights and ensure network security, we updated the Openl Qizhi Community AI Collaboration Platform Usage Agreement in January 2024. The updated agreement specifies that users are prohibited from using intranet penetration tools. After you click "Agree and continue", you can continue to use our services. Thank you for your cooperation and understanding.
For more agreement content, please refer to the《Openl Qizhi Community AI Collaboration Platform Usage Agreement》