11 May 2022 in Scientific Reports
The precise identification of postural behavior plays a crucial role in evaluation of animal welfare and captive management. Deep learning technology has been widely used in automatic behavior recognition of wild and domestic fauna species. The Asian slow loris is a group of small, nocturnal primates with a distinctive locomotion mode, and a large number of individuals were confiscated into captive settings due to illegal trade, making the species an ideal as a model for postural behavior monitoring. Captive animals may suffer from being housed in an inappropriate environment and may display abnormal behavior patterns. Traditional data collection methods are time-consuming and laborious, impeding efforts to improve lorises’ captive welfare and to develop effective reintroduction strategies. This study established the first human-labeled postural behavior dataset of slow lorises and used deep learning technology to recognize postural behavior based on object detection and semantic segmentation. The precision of the classification based on YOLOv5 reached 95.1%. The Dilated Residual Networks (DRN) feature extraction network showed the best performance in semantic segmentation, and the classification accuracy reached 95.2%. The results imply that computer automatic identification of postural behavior may offer advantages in assessing animal activity and can be applied to other nocturnal taxa.
24 May 2024 in IEEE Transactions on Geoscience and Remote Sensing
Semantic segmentation of high spatial resolution remote sensing images presents unique challenges due to the imbalanced foreground-background distribution and large intra-class variance. This study proposes a novel semantic segmentation algorithm based on non-learnable prototypes, named RSProtoSeg. This approach optimizes the spatial relationship between foreground-background prototypes and intra-class prototypes. Specifically, we propose a foreground-background distance optimization loss function to enhance sparsity between these phototypes, effectively mitigating foreground-background distribution imbalances. Moreover, we introduce an online discrete clustering module that represents each class with a set of prototypes and adds an adaptive regular term penalty to promote sparse structure and reduce the variance issue. Evaluation on three remote sensing datasets (iSAID, ISPRS Potsdam, and Vaihingen) demonstrates significant accuracy improvements, aligning our approach with state-of-the-art methods. Our non-learnable prototype-based approach offers a promising solution for semantic segmentation in high spatial resolution remote sensing images.
19 September 2025 in IEEE Transactions on Geoscience and Remote Sensing
High interclass similarity and diverse geometric configurations in high spatial resolution remote sensing images pose significant challenges for remote sensing object detection. To address these challenges, this study proposed a novel remote sensing-oriented object detector MGNet based on multicascaded feature selection and geometric constraints, designed to enhance discriminative feature representation and optimize oriented bounding box regression. First, we designed a novel backbone multicascaded feature selection network (MCFSNet) to improve interclass separability by jointly leveraging and fusing contextual information and multiscale features. Second, we designed a geometric-constrained probabilistic loss, which introduced an angular offset factor and a geometric perception factor to refine angular prediction accuracy of square targets and shape sensitivity. To systematically evaluate angular localization performance, we constructed a multiscale plane remote sensing object detection dataset (MPRSOD) with arbitrary-orientated annotations and proposed the angle error rate (AER), the first dedicated evaluation metric for quantifying angular prediction errors. Extensive experiments on DIOR-R, DOTA V1.5, MAR20, and MPRSOD datasets demonstrate that MGNet obtains competitive mean average precision (mAP) results while maintaining computational efficiency.
23 June 2025 in IEEE Geoscience and Remote Sensing Letters
Challenges in remote sensing object detection (RSOD), such as high interclass similarity, imbalanced foreground–background distribution, and the small size of objects in remote sensing images, significantly hinder detection accuracy. Moreover, the tradeoff between model accuracy and computational complexity poses additional constraints on the application of RSOD algorithms. To address these issues, this study proposes an efficient and lightweight RSOD algorithm integrating multiscale receptive fields and foreground focus mechanism, named robust foreground weighted network (RFWNet). Specifically, we proposed a lightweight backbone network receptive field adaptive selection network (RFASNet), leveraging the rich context information of remote sensing images to enhance class separability. Additionally, we developed a foreground–background separation module (FBSM) consisting of a background redundant information filtering module (BRIFM) and a foreground information enhancement module (FIEM) to emphasize critical regions within images while filtering redundant background information. Finally, we designed a loss function, the weighted CIoU-Wasserstein loss ( $L_{\text {WCW}}$ ), which weights the IoU-based loss by using the normalized Wasserstein distance to mitigate model sensitivity to small object position deviations. The comprehensive experimental results demonstrate that RFWNet achieved 95.3% and 73.2% mean average precision (mAP) with 6.0 M parameters on the DOTA V1.0 and NWPU VHR-10 datasets, respectively, with an inference speed of 52 FPS.
24 September 2025 in IEEE Transactions on Geoscience and Remote Sensing
Unsupervised domain adaptation semantic segmentation for high spatial resolution remote sensing imagery aims to utilize a large amount of labeled data from the source domain to train a supervised model. This model is then adapted for pixel-dense prediction in the unlabeled target domain. Current methods in remote sensing unsupervised domain adaptation mainly focus on mitigating pseudo-label noise. However, they overlook the commonalities between pixel representations across domains. This oversight results in scattered semantic feature representations, thereby limiting segmentation accuracy. To address this issue, this study proposes a novel method prioritizes the extraction of learning universal features, domain-invariant representations spanning low-dimensional discriminative attributes and globally consistent spatial patterns. Specifically, a low-dimensional feature enhancement module is designed to increase feature diversity in shallow layers by employing two learnable scaling and shifting parameters. Additionally, a high-dimensional feature smoothing module uses Gaussian filtering to balance the distribution of global spatial features, enhancing feature learning in deep layers. Comprehensive evaluations on two remote sensing domain adaptation semantic segmentation datasets demonstrate that the superiority of our method, achieving state-of-the-art performance against contemporary baselines in terms of segmentation accuracy and feature alignment.
1 January 2026 in IEEE Transactions on Geoscience and Remote Sensing