Utilizing unsupervised learning, multi-view imaging, and CNN-based attention facilitates cost-effective wetland mapping
Description
The dataset and related code refer to the experiment in the paper of "Utilizing unsupervised learning, multi-view imaging, and CNN-based attention facilitates cost-effective wetland mapping". Due to the storage limit in the Mendeley Data Repository, the data in the >Auto-UNet++>dataset>mvImage>MVfeature directory is not intact. While the data in this directory is only immediate features, all the basic data to implement the Auto-UNet++ from scratch have been fully incorporated in this dataset. Please follow the coding instruction in Appendix E in the paper and go through the demo code to produce the whole data. To ensure successful implementation, please copy the whole dataset to the local disk and do not change the directory structure from the Auto-UNet++ folder (including Auto-UNet++). Please contact us by email, If you encounter any problems while running the demo. Qiao Hu: qiao@huskers.unl.edu