ACM-MM-FRDT

Code repository for our paper entilted "Feature Reintegration over Differential Treatment: A Top-down and Adaptive Fusion Network for RGB-D Salient Object Detection" accepted at ACM MM 2020 (poster).

Overall

Usage Instructions

> Requirment

• Ubuntu 18
• PyTorch 1.3.1
• CUDA 10.1
• Cudnn 7.5.1
• Python 3.7
• Numpy 1.17.3

Train/Test

Before training or testing, please make sure the size of all images is same.

• test
  Download related dataset link and the pretrained model link [fetch code 53x0], and set the param '--phase' as "test" and '--param' as 'True' in demo.py. Meanwhile, you need to set dataset path and checkpoint name correctly.

python demo.py

• train
  Our train-augment dataset link [ fetch code haxl ] , and set the param '--phase' as "train" and '--param' as 'True'(loading checkpoint) or 'False'(no loading checkpoint) in demo.py. Meanwhile, you need to set dataset path and checkpoint name correctly.

python demo.py

If you think this work is helpful, please cite

 @InProceedings{Miao_2020_ACM_MM, 
 author = {Miao {Zhang} and Yu {Zhang} and Yongri {Piao} and Beiqi {Hu} and Huchuan {Lu}},   
          title = {Feature Reintegration over Differential Treatment: A Top-down and Adaptive Fusion Network for RGB-D Salient Object Detection},     
          booktitle = "ACM Multimedia Conference 2020",     
          year = {2020}     
          } 

Comparsion

Results

The results of our method in 7 datasets in here [fetch code t2bx]

Contact Us

If you have any questions, please contact us ( zhangyu4195@mail.dlut.edu.cn ).