In recent years, with the development of deep learning, semantic segmentation for remote sensing images has gradually become a hot issue in computer vision. However, segmentation for multicategory targets is still a difficult problem. To address the issues regarding poor precision and multiple scales in different categories, we propose a UNet, based on multi-attention (MA-UNet). Specifically, we propose a residual encoder, based on a simple attention module, to improve the extraction capability of the backbone for fine-grained features. By using multi-head self-attention for the lowest level feature, the semantic representation of the given feature map is reconstructed, further implementing fine-grained segmentation for different categories of pixels. Then, to address the problem of multiple scales in different categories, we increase the number of down-sampling to subdivide the feature sizes of the target at different scales, and use channel attention and spatial attention in different feature fusion stages, to better fuse the feature information of the target at different scales. We conducted experiments on the WHDLD datasets and DLRSD datasets. The results show that, with multiple visual attention feature enhancements, our method achieves 63.94% mean intersection over union (IOU) on the WHDLD datasets; this result is 4.27% higher than that of UNet, and on the DLRSD datasets, the mean IOU of our methods improves UNet’s 56.17% to 61.90%, while exceeding those of other advanced methods.

Ship detection plays a significant role in military and civil fields. Although some state-of-the-art detection methods, based on convolutional neural networks (CNN) have certain advantages, they still cannot solve the challenge well, including the large size of images, complex scene structure, a large amount of false alarm interference, and inshore ships. This paper proposes a ship detection method from optical remote sensing images, based on visual attention enhanced network. To effectively reduce false alarm in non-ship area and improve the detection efficiency from remote sensing images, we developed a light-weight local candidate scene network( ${\mathrm{L}}^2$ CSN) to extract the local candidate scenes with ships. Then, for the selected local candidate scenes, we propose a ship detection method, based on the visual attention DSOD(VA-DSOD). Here, to enhance the detection performance and positioning accuracy of inshore ships, we both extract semantic features, based on DSOD and embed a visual attention enhanced network in DSOD to extract the visual features. We test the detection method on a large number of typical remote sensing datasets, which consist of Google Earth images and GaoFen-2 images. We regard the state-of-the-art method [sliding window DSOD (SW+DSOD)] as a baseline, which achieves the average precision (AP) of 82.33%. The AP of the proposed method increases by 7.53%. The detection and location performance of our proposed method outperforms the baseline in complex remote sensing scenes.

Few-shot domain adaptive object detection (FSDAOD) aims to transfer knowledge from a source domain to a target domain with limited labeled data, which faces severe challenges in the field of remote sensing. To address this issue, numerous CNN-based domain adaptation methods employ style transfer and feature alignment to mitigate domain shifts, but limited target domain samples are prone to yielding equivocal optimization and are maladaptive. Furthermore, the sparsity of targets and the complexity of backgrounds in remote sensing imagery contribute to confusing feature alignment. Moreover, DETR-based detectors have achieved remarkable progress in unsupervised domain adaptation (UDA) but remain unexplored in FSDAOD. To address these challenges, we introduce FSDA-DETR, the first DETR-based strong baseline designed for the FSDAOD of remote sensing imagery. Specifically, we propose a cross-domain style rectification (CSR) module that rectifies the styles of the target domain to align with the source domain by storing and dynamically updating the weighted-fusion source domain style bases. To further strengthen the detector’s cross-domain detection performance, we propose a category-aware feature alignment (CFA) module that performs fine-grained masking on object regions of rectified backbone features corresponding to different categories and utilizes adversarial training for domain-invariant feature extraction. Extensive experiments on three cross-domain benchmarks, comprising six diverse datasets, demonstrate that FSDA-DETR outperforms state-of-the-art methods. For instance, in the Optical-to-SAR benchmark, FSDA-DETR achieves 74.5% mAP with only 1% of target domain training data. The code and datasets are available at https://github.com/wsybb252237/FSDA-DETR.

Semantic segmentation is crucial for interpreting remote sensing images. The segmentation performance has been significantly improved recently with the development of deep learning. However, complex background samples and small objects greatly increase the challenge of the semantic segmentation task for remote sensing images. To address these challenges, we propose a dual-domain refinement network (DDRNet) for accurate segmentation. Specifically, we first propose a spatial and frequency feature reconstruction (SFFR) module, which separately utilizes the characteristics of the frequency and spatial domains to refine the global salient features and the fine-grained spatial features of objects. This process enhances the foreground saliency and adaptively suppresses background noise. Subsequently, we propose a feature alignment (FA) module that selectively couples the features refined from both domains via cross-attention, achieving semantic alignment between frequency and spatial domains. Additionally, a meticulously designed detail-aware attention (DAA) module is introduced to compensate for the loss of small objects during feature propagation. This module leverages cross-correlation matrices between high-level features and the original image to quantify the similarities among objects belonging to the same category, thereby transmitting rich semantic information from high-level features to small objects. The results on multiple datasets validate that our method outperforms the existing methods and achieves a good compromise between computational overhead and accuracy. Our code is available at https://github.com/Gritiii/DDRNet_real .

Due to strong ocean waves, broken clouds, and extensive cloud cover interferences, ocean ship detection performs poorly when using optical remote sensing images. In addition, it is a challenge to detect small ships on medium resolution optical remote sensing that cover a large area. In this paper, in order to balance the requirements of real-time processing and high accuracy detection, we proposed a novel ship detection framework based on locally oriented scene complexity analysis. First, the proposed method can separate a full image into two types of local scenes (i.e., simple or complex local scenes). Next, simple local scenes would utilize the fast saliency model (FSM) to rapidly complete candidate extraction, and for complex local scenes, the ship feature clustering model (SFCM) will be applied to achieve refined detection against severe background interferences. The FSM considers a fusion enhancement image as an input of the pulse response analysis in the frequency domain to achieve rapid ship detection in simple local scenes. Next, the SFCM builds the descriptive model of the ship feature clustering algorithm to ensure the detection performance on complex local scenes. Extensive experiments on SPOT-5 and GF-2 ocean optical remote sensing images show that the proposed ship detection framework has better performance than the state-of-the-art methods, and it addresses the tricky problem of real-time ocean ship detection under strong waves, broken clouds, extensive cloud cover, and ship fleet interferences. Finally, the proposed ocean ship detection framework is demonstrated on an onboard processing hardware.