Yong Liu

Underwater object detection is crucial for applications such as marine biological resource assessment and ecological monitoring. However, its development is hindered by two major challenges: low signal-to-noise ratios in underwater imagery often cause missed detections of small and camouflaged targets, while high computational costs restrict deployment on edge devices such as autonomous underwater vehicles. To address these issues, we propose EGFR-Net, a lightweight hierarchical edge-guided framework for underwater object detection. First, we design CSP-DenseRepViTBlock, a novel representation enhancement architecture with an adaptive shuffle-fuse mechanism, which improves feature robustness while maintaining a lightweight design. Second, we propose an Edge-Guided Attention (EGA) module to extract robust edge priors for boundary-aware feature learning. Finally, we construct a hierarchical edge-guidance framework, where an Edge-Guide Module (EGM) generates dedicated edge-guidance signals and a Guided Feature Fusion Module (GFFM) hierarchically integrates these signals at multiple stages to refine semantic features. Experimental results show that, compared with YOLOv12L, EGFR-Net (α = 0.9) improves mAP50 by 0.4%, 0.5%, and 0.1%, and mAP50-95 by 0.6%, 0.5%, and 0.9% on the UTDAC2020, URPC2020, and DUO datasets, respectively, while reducing parameters by 27% and computational cost by 36%. Moreover, compared with YOLOv12L, the small version, EGFR-Net (α = 0.5), achieves comparable performance using about 50% of the parameters and 47% of the computational cost.

Remote sensing (RS) image segmentation faces persistent challenges due to limited labeled data and the presence of long-tailed distribution. While semi-supervised learning (SSL) can leverage unlabeled data, it often struggles to address the severe class imbalance that manifests as a long-tailed distribution. To tackle this, we propose the Dynamic Category Preference-Aware Semantic Segmentation Framework (CPSeg). Unlike existing methods that estimate model’s category preference on pseudo-labels or rely on prior information, CPSeg introduces a simple yet effective approach to directly model this preference. By feeding a Pattern-Free Image (PFI) into the model, we capture its current tendencies toward certain categories without any prior knowledge. This dynamically learned category preference drives two key components of CPSeg. First, it enables adaptive weight allocation, which adjusts the training focus by assigning higher weights to low-frequency categories, promoting balanced learning across all categories. Second, CPSeg introduces a Category-wise Pairwise Memory Bank (CPMB) and integrates category preference information to perform dynamic preference-aware data augmentation, making the augmentation process more suitable for addressing the long-tailed distribution problem. When compared with other advanced approaches, CPSeg exhibits superior performance, achieving mean intersection over union (mIoU) scores of 74.50% on the GID-15 and 59.26% on the MSl dataset, respectively, under a challenging 1/8 labeled data setting, highlighting its strong potential for addressing long-tailed distribution and class imbalance in RS segmentation.

Marine benthic organism detection is a critical field in underwater object detection. Due to the small size, diverse morphology, and occlusion of marine benthic organisms, general object detection algorithms struggle to achieve high accuracy in this domain. To address these challenges, we propose the multi-scale feature extraction and reinforced bounding box regression for the marine benthos detection (MBR-YOLO) algorithm. The key innovations of MBR-YOLO are as follows: First, we introduce the cross-stage multi-scale aggregation inception module, which uses multi-scale convolutions to extract features and parallel branches to fuse contextual information. This enhances the model’s sensitivity to small object features and mitigates the problem of insufficient feature extraction for small targets. Second, we propose a three-dimensional group convolution channel shuffle feature pyramid network, which combines features of different scales. This structure employs a feature extraction mechanism to automatically capture scale information and leverages channel transformation to reorganize multi-scale details. This improves the model’s ability to perceive multi-scale targets and maintain invariance in fine details. Finally, we design minimal geometry distance Intersection over Union (MGDIoU) and MGDIoU-non-maximum suppression to address occlusion issues. MGDIoU redefines the minimum geometric diagonal distance metric, enhancing Intersection over Union computation. Using MGDIoU as a guiding strategy for non-maximum suppression preserves more predictions closer to the ground truth, reducing missed detections. Experiments demonstrate that MBR-YOLO improves mAP0.5:0.95 by 4.8%, 5.5%, and 5.5% on the Underwater Robot Picking Contest 2020 (URPC2020), SUODAC, and AUDD datasets, respectively. Compared with 13 state-of-the-art detection algorithms, including YOLOv10, MBR-YOLO achieves higher accuracy. MBR-YOLO provides robust data support for marine benthic detection and offers a theoretical reference for future research in related fields.

PurposeIn recent years, the continuous advancement of convolutional neural networks (CNNs) has led to the widespread integration of deep neural networks as a mainstream approach in clinical diagnostic support. Particularly, the utilization of CNN-based medical image segmentation has delivered favorable outcomes for aiding clinical diagnosis. Within this realm, network architectures based on the U-shaped structure and incorporating skip connections, along with their diverse derivatives, have gained extensive utilization across various medical image segmentation tasks. Nonetheless, two primary challenges persist. First, certain organs or tissues present considerable complexity, substantial morphological variations, and size discrepancies, posing significant challenges for achieving highly accurate segmentation. Second, the predominant focus of current deep neural networks on single-resolution feature extraction limits the effective extraction of feature information from complex medical images, thereby contributing to information loss via continuous pooling operations and contextual information interaction constraints within the U-shaped structure.ApproachWe proposed a five-layer pyramid segmentation network (PS5-Net), a multiscale segmentation network with diverse resolutions that is founded on the U-Net architecture. Initially, this network effectively leverages the distinct features of images at varying resolutions across different dimensions, departing from prior single-resolution feature extraction methods to adapt to intricate and variable segmentation scenarios. Subsequently, to comprehensively integrate feature information from diverse resolutions, a kernel selection module is proposed to assign weights to features across different dimensions, enhancing the fusion of feature information from various resolutions. Within the feature extraction network denoted as PS-UNet, we preserve the classical structure of the traditional U-Net while enhancing it through the incorporation of dilated convolutions.ResultsPS5-Net attains a Dice score of 0.9613 for liver segmentation on the CHLISC dataset and 0.8587 on the ISIC2018 dataset for skin lesion segmentation. Comparative analysis with diverse medical image segmentation methodologies in recent years reveals that PS5-Net has achieved the highest scores and substantial advancements.ConclusionsPS5-Net effectively harnesses the rich semantic information available at different resolutions, facilitating a comprehensive and nuanced understanding of the input medical images. By capitalizing on global contextual connections, the network adeptly captures the intricate interplay of features and dependencies across the entire image, resulting in more accurate and robust segmentation outcomes. The experimental validation of PS5-Net underscores its superior performance in medical image segmentation tasks, offering promising prospects for enhancing diagnostic and analytical processes within clinical settings. These results highlight the potential of PS5-Net to significantly contribute to the advancement of medical imaging technologies and ultimately improve patient care through more precise and reliable image analysis.