SAR Target Incremental Recognition Based on Hybrid Loss Function and Class-Bias Correction

Yongsheng Zhou,Shuo Zhang,Xiaokun Sun,Fei Ma,Fan Zhang

doi:10.3390/app12031279

Yongsheng Zhou, Shuo Zhang + Show 3 more

Open Access

https://doi.org/10.3390/app12031279

Copy DOI

Journal: Applied sciences	Publication Date: Jan 25, 2022
Citations: 8	License type: CC BY 4.0

Affiliation: Beijing University of Chemical Technology

Abstract

The Synthetic Aperture Radar (SAR) target recognition model usually needs to be retrained with all the samples when there are new-coming samples of new targets. Incremental learning emerges to continuously obtain new knowledge from new data while preserving most previously learned knowledge, saving both time and storage. There are still three problems in the existing incremental learning methods: (1) the recognition performance of old target classes degrades significantly during the incremental process; (2) the target classes are easily confused when similar target classes increase; (3) the model is inclined to new target classes due to class imbalance. Regarding the three problems, firstly, the old sample preservation and knowledge distillation were introduced to preserve both old representative knowledge and knowledge structure. Secondly, a class separation loss function was designed to reduce the intra-class distance and increase the inter-class distance, effectively avoiding the confusion between old and new classes. Thirdly, a bias correction layer and a linear model was designed, which enabled the model to treat the old and new target classes more fairly and eliminate the bias. The experimental results on the MSTAR dataset verified the superior performance compared with the other incremental learning methods.

Highlights

Remote sensing is a detection technology that obtains information about objects without direct contact at long distances
The method in this paper addresses the catastrophic forgetting problem, recognition confusion problem, and bias problem caused by class imbalance in Synthetic Aperture Radar (SAR) target incremental learning
The incremental process can be divided into 2-phases, 5-phases, and 10-phases increments depending on the number of new target classes each time. 5-phases increment means the process of initial training on 2 classes of old data, followed by increments of 2 classes of new data each time

Summary

Introduction

Remote sensing is a detection technology that obtains information about objects without direct contact at long distances. It plays a very important role in many studies, and an increasing number of more precise sensors and measurements are providing researchers with more information [1]. It has a growing impact in a wide variety of areas from business to science to public policy. It is usually challenging to collect the labeled samples of all the target classes at once in practice, and the training data of the new target classes are obtained gradually. Due to the limitation of computation and storage resources in some particular application situations, this approach cannot be applied

Methods

Results

Conclusion