Ensemble learning for LCLU classification
This repository contains the code for the paper titled "Deep Neural Network Ensembles for Remote Sensing Land Cover and Land Use Classification"
Stochastic Weight Averaging (Izmailov et al., 2019), Snapshot Ensemble (Huang et al., 2017), and Fast Geometric Ensemble (Garipov et al., 2018) techniques are used to check whether adopting ensemble techniques improve the LCLU classification performance. Experiments conducted on NWPU-RESISC45 (Cheng et al., 2017) and AID (Xia et al., 2017) datasets help to point the best performing ensemble technique.
Citation
Please kindly cite the paper below if this code is useful and helpful for your research.
TBD
Contact Information:
If you encounter bugs while using this code, please do not hesitate to contact me.
Burak Ekim: ekim19@itu.edu.tr