Awesome

MEANet

Pytorch implementation for MEANet: Multi-Modal Edge-Aware Network for Light Field Salient Object Detection.

Requirements

• Python 3.6 <br>
• Torch 1.10.2 <br>
• Torchvision 0.4.0 <br>
• Cuda 10.0 <br>
• Tensorboard 2.7.0

Usage

To Train

• Download the training dataset and modify the 'train_data_path'.
• Start to train with

python -m torch.distributed.launch --nproc_per_node=4 train.py 

To Test

• Download the testing dataset and have it in the 'dataset/test/' folder.
• Download the already-trained MEANet model and have it in the 'trained_weight/' folder.
• Change the weight_name in test.py to the model to be evaluated.
• Start to test with

python test.py  

Download

Trained model for testing

We released two versions of the trained model:

Trained with additional 100 samples from HFUT-Lytro on baidu pan with fetch code: 0o0r or on Google drive

Trained only with DUTLF-FS on baidu pan with fetch code: 75bn or on Google drive

Saliency map

We released two versions of the saliency map:

Trained with additional 100 samples from HFUT-Lytro on baidu pan with fetch code: x7xa or on Google drive

Trained only with DUTLF-FS on baidu pan with fetch code: s7vn or on Google drive

Citation

Please cite our paper if you find the work useful:

@article{JIANG202278,
title = {MEANet: Multi-modal edge-aware network for light field salient object detection},
journal = {Neurocomputing},
volume = {491},
pages = {78-90},
year = {2022},
author = {Yao Jiang and Wenbo Zhang and Keren Fu and Qijun Zhao}
}