Abstract
Scene classification is one of the fundamental techniques shared by many basic remote sensing tasks with a wide range of applications. As the demands of catering with situations under high variance in the data urgent conditions are rising, a research topic called few-shot scene classification is receiving more interest with a focus on building classification model from few training samples. Currently, methods using the meta-learning principle or graphical models are achieving state-of-art performances. However, there are still significant gaps in between the few-shot methods and the traditionally trained ones, as there are implicit data isolations in standard meta-learning procedure and less-flexibility in the static graph neural network modeling technique, which largely limit the data-to-knowledge transition efficiency. To address these issues, this paper proposed an novel few-shot scene classification algorithm based on a different meta-learning principle called continual meta-learning, which enhances the inter-task correlation by fusing more historical prior knowledge from a sequence of tasks within sections of meta-training or meta-testing periods. Moreover, as to increase the discriminative power between classes, a graph transformer is introduced to produce the structural attention, which can optimize the distribution of sample features in the embedded space and promotes the overall classification capability of the model. The advantages of our proposed algorithm are verified by comparing with nine state-of-art meta-learning based on few-shot scene classification on three popular datasets, where a minimum of a 9% increase in accuracy can be observed. Furthermore, the efficiency of the newly added modular modifications have also be verified by comparing to the continual meta-learning baseline.
Highlights
The first section will briefly introduce datasets UC Merced Landuse [78], Northwestern Polytechnical University (NWPU)-RESISC45 [79], and Aerial Image Dataset (AID) [80], which are used for training and evaluation, which are commonly used datasets with increasing complexities that ensure the objective analysis of the behavior of the model
Based on the comparisons with counterparts and baseline, as the continual meta learning framework makes a better utilization of the historical prior, the averaged classification accuracies of our proposed and baseline algorithms stay almost always among the top 2 positions when compared to counterparts
Few-shot remote sensing scene classification is one such fundamental technique that is indispensable in a large variety of tasks
Summary
Many efforts have been made to alleviate these two issues under the traditional training framework, common methods include using the correct model parameter transfer methods to reduce the feature shift in cross domain model application [11], or to augment a dataset with artificial samples generated by generative adversarial network techniques [10] Due to these problems, a new research topic for model training under data urgent conditions called Few-Shot Learning (FSL) has come into the view. Some methods utilized the heatedly studied attention mechanism such as attention fusion [31,32] and attention metric calculation [33] to improve discriminative power Among these FSL-based remote sensing scene classification methods, one category based on meta-learning has received much more attention in recent researches [34,35]. We further utilize structural attention to improve the node feature encoding via the graph transformer [58], which is further combined with an edge labeling Bayesian graph network [52] to further increase the categorical discriminative power
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
