Skip to content

Latest commit

 

History

History
48 lines (38 loc) · 3.99 KB

README.md

File metadata and controls

48 lines (38 loc) · 3.99 KB
Raw

SOSMaskFuse: An Infrared and Visible Image Fusion Architecture Based on Salient Object Segmentation Mask

G. Li, X. Qian and X. Qu, "SOSMaskFuse: An Infrared and Visible Image Fusion Architecture Based on Salient Object Segmentation Mask," in IEEE Transactions on Intelligent Transportation Systems, doi: 10.1109/TITS.2023.3268063.

Platform

Python 3.7
Pytorch 1.7.1

For test

  • Model

Make sure the two models provided are on the correct path.

  1. The training model "Epoch_model_1.model" of the Autoencoder network. Please will be placed in the model ./SOSnetwork/IMV_F_Autoencoder/models/sosmaskfuse_autoencoder/1e2/Epoch_model/.
  2. Training model of Autoencoder network (Verification code: l22i.). Please will be placed in the model ./SOSnetwork/model/.

  • Test images

During the testing of SOSMaskFuse, the folder storing infrared images is named "Inf" and the folder storing visible is named "Vis". (The files in the SOSmask folder are binary significance masks of infrared images generated by the SOSnetwork model.) The corresponding infrared image and visible image file name must be the same. Please place infrared and visible images in the following directory: ./SOSnetwork/Test_images/InfVis/.

For train

IMV-F_Autoencoder network

  • Dataset

MS-COCO 2014 is utilized to train our IMV-F_Autoencoder network. Please place images in the following directory: ./IMV_F_Autoencoder/train_dataset/Disk_B/MSCOCO2014/train2014/. [T.-Y. Lin, M. Maire, S. Belongie, J. Hays, P. Perona, D. Ramanan, P. Dollar, and C. L. Zitnick. Microsoft coco: Common objects in context. In ECCV, 2014. 3-5.]

SOSnetwork

  • Dataset

ROAD (Verification code: 3txt.) is utilized to train our SOSnetwork network. During the training of SOS model, the folder storing infrared images is named "Inf" and the folder storing GroundTruth is named "GT". The corresponding infrared image and GT file name must be the same. Put the above folder in the following directory: ./SOSnetwork/data/road/. [Takumi, Karasawa, Kohei Watanabe, Qishen Ha, Antonio Tejero-De-Pablos, Yoshitaka Ushiku, and Tatsuya Harada. "Multispectral object detection for autonomous vehicles." In Proceedings of the on Thematic Workshops of ACM Multimedia 2017, pp. 35-43. 2017.]

  • Model

hrnetv2_w30_imagenet_pretrained (Verification code: muau.) serves as a pre-training model for training SOS network. Please will be placed in the model ./SOSnetwork/model/pretrain_model/. [Wang J, Sun K, Cheng T, Jiang B, Deng C, Zhao Y, Liu D, Mu Y, Tan M, Wang X, Liu W. Deep high-resolution representation learning for visual recognition. IEEE transactions on pattern analysis and machine intelligence. 2020 Apr 1;43(10):3349-64.]

If you have any question about this code, feel free to reach me([email protected])

The method is briefly introduced in Chinese in the published blog. → CSDN blog link

@ARTICLE{10109138,
  author={Li, Guofa and Qian, Xuanhu and Qu, Xingda},
  journal={IEEE Transactions on Intelligent Transportation Systems}, 
  title={SOSMaskFuse: An Infrared and Visible Image Fusion Architecture Based on Salient Object Segmentation Mask}, 
  year={2023},
  volume={},
  number={},
  pages={1-20},
  doi={10.1109/TITS.2023.3268063}}