Lightweight Channel Attention and Multiscale Feature Fusion Discrimination for Remote Sensing Scene Classification

Abstract

High-resolution remote sensing image scene classification has attracted widespread attention as a basic earth observation task. Remote sensing scene classification aims to assign specific semantic labels to remote sensing scene images to serve specified applications. Convolutional neural networks are widely used for remote sensing image classification due to their powerful feature extraction capabilities. However, the existing methods have not overcome the difficulties of large-scene remote sensing images of large intraclass diversity and high interclass similarity, resulting in low performance. Therefore, we propose a new remote sensing scene classification method that combines lightweight channel attention and multiscale feature fusion discrimination, called LmNet. First, ResNeXt is used as the backbone; second, a new lightweight channel attention mechanism is constructed to quickly and adaptively learn the salient features of important channels. Furthermore, we designed a multiscale feature fusion discrimination framework, which fully integrates shallow edge feature information and deep semantic information to enhance feature representation capabilities and uses multiscale features for joint discrimination. Finally, a cross-entropy loss function based on label smoothing is built to reduce the influence of interclass similarity on feature representation. In particular, our lightweight channel attention and multiscale feature fusion mechanism can be flexibly embedded in any advanced backbone as a functional module. The experimental results on three large-scale remote sensing scene classification datasets show that compared with the existing advanced methods, our proposed high-efficiency end-to-end scene classification method has reached state-of-the-art. Moreover, our method has a weaker dependence on labeled data and provided better generalization performance.INDEX TERMS Remote sensing scene classification, convolutional neural network, lightweight channel attention, multiscale feature fusion, label smoothing.

Multi-scale information fusion network with label smoothing strategy for corneal ulcer classification in slit lamp images

“…In addition, compared with EfficientNetB2 with comparable model parameters, our proposed method has also made great improvement in terms of all evaluation metrics. It is worth noting that the proposed method is also compared with a recent study on multi-scale feature fusion and label smoothing, which is proposed for remote sensing classification and is named LmNet ( Wan et al, 2021 ). LmNet takes pre-trained ResNext50 as backbone and combines channel attention, multi-scale feature fusion and label smoothing.…”

“…This characteristic encourages the model to learn in the direction with the greatest difference between the correct label and the wrong label, which means only the loss of the correct label position is calculated in the optimization process of the model. However, when the training data is small, and the inter-class similarity and intra-class differences is relatively large, it may cause the network to be overfitting ( Wan et al, 2021 ). To solve the above problems and inspired by previous study ( Szegedy et al, 2017 ; Müller et al, 2019 ), label smoothing is introduced in this study.…”

“…Previous studies have shown that some shallow features can improve classification performance ( Lee et al, 2015 ; Wang et al, 2015 ; Ma et al, 2019 ), which has attracted our attention. In addition, considering that the misclassification of corneal ulcers may be caused by images of similar but different categories in the dataset ( Wan et al, 2021 ), we introduce label smoothing in the cross-entropy loss, which can reduce the inter-class similarity and intra-class differences ( Müller et al, 2019 ). In short, we develop a multi-scale information fusion network with label smoothing strategy to achieve the classification of corneal ulcer.…”

Scene Classification of Remote Sensing Images Using EfficientNetV2 with Coordinate Attention

“…Xu R et al [12] combined two different attention mechanisms, control gate attention and feedback attention, in the main and non-main positions of the network, experiment to prove the validity of the model. Wan H et al [13] designed a multi-scale fusion discrimination framework, which is used to add a lightweight attention mechanism to quickly learn important features in the channel and the ability to dilute edge information, and has a good generalization ability. Wu H et al [14] proposed a self-attention network model with joint loss, using the attention model proposed by ResNet-18 integrated to extract image features and reduce the interference of redundant information in complex images.…”

Transformer based on channel-spatial attention for accurate classification of scenes in remote sensing image

“…Classification performance largely depends on features that accurately represent the scene in the image, and thus the extraction of features that describe an image more accurately has become a primary research focus. Recently, convolutional neural networks (CNNs) have been widely utilized in scene classification because they are capable of extracting high-level semantic feature representations for scene classification 13 – 15 . However, the spatial relationships between features in HRRS images are complex, and there is a large amount of redundant information.…”