This research aims to enhance the effectiveness of image classification using semi-supervised learning techniques in PolSAR images.
Utilized a hybrid convolutional neural network (CNN) approach for image processing.
Implemented crosspseudo supervision for refining classification accuracy in the dataset.
Conducted comparisons with traditional labeled datasets to evaluate performance.
Significant improvement in classification accuracy compared to traditional methods.
Achieved effective performance even with reduced labeled data availability.
Abstract
In computer vision applications, convolutional neural networks(CNNs) have demonstrated significant effectiveness and achievedremarkable success, largely because of the abundance of well-labelleddatasets.