This is an implementation of "PolSAR Image Classification using a Hybrid Complex-Valued Network (HybridCVNet)" Accepted for Publication on IEEE-GRSL. The paper can be accessed through: https://ieeexplore.ieee.org/document/10693615
Two Commonly used datasets were used in this paper, namely Flevoland and San Francisco Flevoland dataset can be downloaded from: https://github.com/mqalkhatib/SDF2Net/tree/main/Datasets/Flevoland
Python 3.9.18, Tensorflow (and Keras) 2.10.0, cvnn 2.0, Tensorflow Probability 0.18.0
To quantitatively measure the proposed HybridCVNet model, three evaluation metrics are employed to verify the effectiveness of the algorithm, Overall Accuracy (OA), Average Accuracy (AA) and Cohen's Kappa (k). Also, Each class accuracy has been reported
Model was qualitatively evaluated by visually comparing the resulting class maps.
@ARTICLE{10693615, author={Alkhatib, Mohammed Q.}, journal={IEEE Geoscience and Remote Sensing Letters}, title={PolSAR Image Classification Using a Hybrid Complex-Valued Network (HybridCVNet)}, year={2024}, volume={21}, number={}, pages={1-5}}
Feel free to contact me on: mqalkhatib@ieee.org