Object detection in complex optical remote sensing images is a challenging problem due to the wide variety of scales, densities, and shapes of object instances on the earth surface. In this letter, we focus on the wide-scale variation problem of multiclass object detection and propose an effective object detection framework in remote sensing images based on YOLOv2. To make the model adaptable to multiscale object detection, we design a network that concatenates feature maps from layers of different depths and adopt a feature introducing strategy based on oriented response dilated convolution. Through this strategy, the performance for small-scale object detection is improved without losing the performance for large-scale object detection. Compared to YOLOv2, the performance of the proposed framework tested in the DOTA (a large-scale data set for object detection in aerial images) data set improves by 4.4% mean average precision without adding extra parameters. The proposed framework achieves real-time detection for $1024\times 1024$ image using Titan Xp GPU acceleration. ^{1 1}https://github.com/WenchaoliuMUC/Detection-of-Multiclass-Objects-in-Optical-Remote-Sensing-Images

In recent times, many efforts have been made to improve remote sensing image scene classification, especially using popular deep convolutional neural networks. However, most of these methods do not consider the specific scene orientation of the remote sensing images. In this letter, we propose the improved oriented response network (IORN), which is based on the ORN, to handle the orientation problem in remote sensing image scene classification. We propose average active rotating filters (A-ARFs) in the IORN. While IORNs are being trained, A-ARFs are updated by a method that is different from the ARFs of the ORN, without additional computations. This change helps IORN improve its ability to encode orientation information and speeds up optimization during training. We also propose Squeeze-ORAlign (S-ORAlign) by adding a squeeze layer to ORAlign of ORN. With the squeeze layer, S-ORAlign can address large-scale images, unlike ORAlign. An ablation study and comparison experiments are designed on a public remote sensing image scene classification data set. The experimental results demonstrate the effectiveness and better performance of the proposed model over that of other state-of-the-art models.

Change detection (CD) is a challenging task on high-resolution bitemporal remote sensing images. Many recent studies of CD have focused on designing fully convolutional Siamese network architectures. However, most of these methods initialize their encoders by random values or an ImageNet pretrained model, without any prior for the CD task, thus limiting the performance of the CD model. In this paper, the novel supervised contrastive pretraining and fine-tuning change detection (SCPFCD) framework, which is made up of two cascaded stages, is presented to train a CD network based on a pretrained encoder. In the first supervised contrastive pretraining stage, the encoder of the Siamese network is asked to solve a joint pretext task introduced by the proposed CDContrast pretraining method on labeled CD data. The proposed CDContrast pretraining method includes land contrastive learning (LCL), which is based on supervised contrastive learning, and proxy CD learning. The LCL focuses on learning the spatial relationships among the land cover from bitemporal images by solving a land contrast task, while the proxy CD learning performs a proxy CD task on the top of the upsampling projector, to avoid local optima for the LCL and learn features for the CD. Then, in the second fine-tuning stage, the whole Siamese network initialized with the pretrained encoder is fine-tuned to perform the CD task in an end-to-end manner. The proposed SCPFCD framework was verified with three CD datasets of high-resolution remote sensing images. The extensive experimental results consistently show that the proposed framework can effectively improve the ability to extract change information for Siamese networks.

Deep learning (DL) has been introduced to change detection (CD) due to its powerful feature representation and robust generalization abilities. However, the application of large DL models has high computational complexity and massive storage requirements for achieving good performance. For disaster emergency response and other applications with high timeliness requirements, it is difficult to deploy large DL models on spaceborne edge devices with limited resources to achieve on-board CD processing. To address this limitation, a novel CD based on knowledge distillation (CDKD) method that combines prototypical contrastive (PC) distillation and channel-spatial-normalized (CSN) distillation is proposed. PC distillation represents the feature distribution by calculating the differences between the similarities of pixel features and their positive and negative prototypes, and improves the student model's detection ability in changed regions that have similar features to the background by mimicking the relative feature distribution. CSN distillation combines two distillation paradigms, channel normalization and spatial normalization, and guides the student model to comprehensively learn the knowledge contained in the output probabilities of the teacher model to accurately identify changed regions with complex shapes. The effectiveness and reliability of the proposed CDKD method are verified on three public remote sensing CD datasets, and extensive experiments and analyses show that the proposed CDKD method can be used to train lightweight models with comparable performance to that of large models.

SAR-optical data fusion cloud removal is a highly promising cloud removal technology that has attracted considerable attention. In this field, primary researches are based on deep learning, which can be divided into two categories: convolutional neural network (CNN)-based and Transformer-based. In cases of extensive cloud coverage, CNN-based methods, with their local spatial awareness, effectively capture local structural information in SAR data, preserving clear contours of cloud-removed land covers. However, these methods struggle to capture global land cover information in optical images, often resulting in notable color discrepancies between recovered and cloud-free regions. Conversely, Transformer-based methods, with their global modeling capability and inherent low-pass filtering properties, excel at capturing long-range spatial correlations in optical images, thereby maintaining color consistency across cloud-removal outputs. However, these methods are less effective at capturing the fine structural details in SAR data, which can lead to blurred local contours in the final cloud-removed images. In this context, a novel framework called heterogeneous parallel network for cloud removal (HPN-CR) is proposed to achieve high-quality cloud removal under extensive cloud coverage conditions. HPN-CR employs the proposed heterogeneous encoder with its SAR-optical input approach to effectively extract and fuse the local structural information in cloudy areas from SAR images, with the spectral information about land covers in cloud-free areas from the whole optical images. In particular, it uses a ResNet network with local spatial awareness to extract SAR features. It also uses the proposed Decloudformer, which globally models multi-scale spatial correlations, to extract optical features. The output features are fused by the heterogeneous encoder and then reconstructed to cloud-removal images through a pixelshuffle-based decoder. Comprehensive experiments were conducted and experimental results demonstrated the effectiveness and superiority of the proposed method. The code is available at https://github.com/G-pz/HPN-CR.

Recently, remote sensing image scene classification technology has been widely applied in many applicable industries. As a result, several remote sensing image scene classification frameworks have been proposed; in particular, those based on deep convolutional neural networks have received considerable attention. However, most of these methods have performance limitations when analyzing images with large intraclass variations. To overcome this limitation, this letter presents the marginal center loss with an adaptive margin. The marginal center loss separates hard samples and enhances the contributions of hard samples to minimize the variations in features of the same class. Experimental results on public remote sensing image scene data sets demonstrate the effectiveness of our method. After the model is trained using the marginal center loss, the variations in the features of the same class are reduced. Furthermore, a comparison with state-of-the-art methods proves that our model has competitive performance in the field of remote sensing image scene classification.