CPNet

Consistency perception network for 360° omnidirectional salient object detection

This repo is an official implementation of the CPNet, which has been accepted by the journal Neurocomputing, 2025.

Prerequisites

• python=3.x
• pytorch=1.0.0+
• torchvision
• numpy
• opencv-python

Usage

1. Clone the repository

git clone https://github.com/Brook-Wen/CPNet.git
cd CPNet/

2. Training

python main.py --mode='train'

3. Testing

python main.py --mode='test' --batch_size 1 --model='[YOUR PATH]' --test_fold='[SAVE PATH]' --sal_mode='[DATASET]'

• We provide the trained model weights (fetch code: fqft).

4. Evaluation

• We provide the predicted saliency maps (fetch code: xepc) of our CPNet on three datasets.
• You can use this toolbox for evaluation.

Citation

If you think this work is helpful, please cite

@inproceedings{wen2025cpnet,
  title={Consistency perception network for 360° omnidirectional salient object detection},
  author={Wen, Hongfa and Zhu, Zunjie and Zhou, Xiaofei and Zhang, Jiyong and Yan, Chenggang},
  booktitle={Neurocomputing},
  year={2025}
}

• If you have any questions, feel free to contact me via: hf_wen(at)outlook.com.