The underwater target detection is the most important part of monitoring for environment, ocean, and other fields. However, the detection accuracy is greatly decreased by the poor image quality resulted from the complex underwater environments. The storage and computing power of underwater equipments are not enough for complex underwater target detection technology. Therefore, many YOLO series algorithms have been applied to underwater target detection. On the basis of YOLOv8, a lightweight underwater detector enhanced by Attention mechanism, GSConv and WIoU, AGW-YOLOv8, is proposed in this paper. Firstly, by the combination of limited contrast adaptive histogram equalization and wavelet transform(LCAHE-WT), the fidelity and detail of images are improved; Secondly, by CBAM, the key channel features can be effectively extracted with retaining spatial information to improve the performance of the network when dealing with complex image tasks; Thirdly, by GSConv, composed of depth-wise separable convolution and regular convolution, the model parameters and computational complexity are reduced; Fourth, the SE attention mechanism is integrated into the C2f module of the neck, and the channel dimension is weighted to make the network more focus on important features, and to further enhance the feature extraction capability; Finally, by the dynamic nonmonotonic mechanism of WIoU, the gradient gain can be reasonably distributed, the harmful gradients of extreme samples can be reduced, and the generalization ability and overall performance of the model are improved. By the experiments on the URPC2020 data-set, it can been proved that the mAP of AGW-YOLOv8 can reaches 82.9%, is 2.5% higher than that of YOLOv8; and the parameters is 2.95M, is lower than 3.01M of YOLOv8.

Underwater object detection is of great significance to marine ecosystems and underwater biodiversity. However, uneven lighting, color distortion, and noise interference in underwater environments severely impact image quality, significantly reducing detection robustness. With limited computational power and storage space, underwater equipment often cannot meet the demands for efficient processing. As a result, the YOLO algorithm has been widely applied in underwater object detection. This paper proposes a lightweight underwater detector, MAW-YOLOv11, based on multi-scale edge information selection. First, dark channel prior is used to estimate the fog concentration in the image, restoring image clarity and enhancing the recognizability of the target. Next, an innovative Multi-Scale Edge Information Select (MSEIS) module is proposed, and based on MSEIS, the C3kMSEIS module is subsequently introduced. These modules are individually incorporated into the C3K2 module of the backbone network, with the aim of extracting features at multiple scales, emphasizing edge information, and efficiently selecting key features that are highly relevant to the target task, thereby improving the model’s accuracy in recognizing critical targets. Third, the ADown downsampling structure is employed to significantly reduce computational overhead while maintaining the original precision. Finally, by using the boundary box regression loss WIoUv3 based on the dynamic focusing mechanism as the loss function, the model’s attention to low-quality samples is increased, effectively reducing the negative impact of these samples on the model’s performance. Experimental results show that on the URPC dataset, the mAP of MAW-YOLOv11 reaches 81.4%, an improvement of 2.1% over YOLOv11, with a parameter count of 2.11M, a reduction of 0.47M compared to YOLOv11. Comparative experiments with other mainstream object detection algorithms validate the effectiveness and superiority of the proposed method.

Deep neural networks are used to accurately detect, estimate, and predict human body poses in images or videos through deep learning-based human pose estimation. However, traditional multi-person pose estimation methods face challenges due to partial occlusions and overlaps between multiple human bodies and body parts. To address these issues, we propose EE-YOLOv8, a human pose estimation network based on the YOLOv8 framework, which integrates Efficient Multi-scale Receptive Field (EMRF) and Expanded Feature Pyramid Network (EFPN). First, the EMRF module is employed to further enhance the model’s feature representation capability. Second, the EFPN optimizes cross-level information exchange and improves multi-scale data integration. Finally, Wise-IoU replaces the traditional Intersection over Union (IoU) to improve detection accuracy through precise overlap measurement between predicted and ground-truth bounding boxes. We evaluate EE-YOLOv8 on the MS COCO 2017 dataset. Compared to YOLOv8-Pose, EE-YOLOv8 achieves an AP of 89.0% at an IoU threshold of 0.5 (an improvement of 3.3%) and an AP of 65.6% over the IoU range of 0.5–0.95 (an improvement of 5.8%). Therefore, EE-YOLOv8 achieves the highest accuracy while maintaining the lowest parameter count and computational complexity among all analyzed algorithms. These results demonstrate that EE-YOLOv8 exhibits superior competitiveness compared to other mainstream methods.

The safety and reliability of rotating machinery hinge significantly on the proper functioning of rolling bearings. In the last few years, there have been significant advances in the algorithms for intelligent fault diagnosis of bearings. However, the vibration signals collected by machines are inevitably affected by irrelevant noise because of the complex working environments of bearings. So, an end-to-end bearing fault diagnosis method: GMSCNN, a bearing fault diagnosis method based on Gram Matrix (GM) and Multi scale Convolutional Neural Network (MSCNN), is proposed in this paper. In this method, first, GM is used to reduce the noise of the collected vibration signals; Secondly, MSCNN is used for feature extraction, and the characteristics of vibration signals at different frequencies and time scales can be captured by the convolutional kernels of different scales; thirdly, two feature enhancement branches are added, utilizing the undenoised vibration signal as input, to enrich and diversify features while enhancing the model’s expressive and generalization capabilities; Finally, the experimental analysis was conducted on two bearing datasets to indicates that the noise robustness of GMSCNN is strong.

Underwater object detection is crucial for marine engineering yet challenged by complex optical properties causing blurring, low contrast, and texture loss. Furthermore, deploying deep learning models on resource-constrained platforms necessitates balancing accuracy with latency. We propose a novel lightweight framework based on the Real-Time Detection Transformer (RT-DETR). First, we design the FasterFDBlock backbone, integrating Partial Convolution with Frequency-domain Dynamic Convolution. This utilizes frequency band modulation to adaptively suppress high-frequency noise and enhance edge details, optimizing feature extraction with reduced redundancy. Second, to mitigate small object detail loss, we introduce the AIFI-SEFN encoder, incorporating a Spatially Enhanced Feed-Forward Network to synthesize global semantic context with local spatial data. Third, a Multi-scale Feature Modulation (MFM) module is applied to dynamically weight deep semantic and shallow detailed features, bolstering robustness against scale variations and background interference. Experimental results on the UTDAC2020 dataset show our method achieves a mean Average Precision (mAP) of 72.1%, outperforming the baseline by 1.7%. Crucially, parameters and Floating Point Operations (GFLOPs) are reduced by 27.1% and 24.6%, respectively. With an inference speed of 72.6 FPS, this model offers a highly efficient solution for real-time underwater perception.

Bearing fault diagnosis has attracted increasing attention due to its critical role in monitoring the health of rotating machinery. Data-driven models based on deep learning (DL) have demonstrated strong capabilities in feature extraction. However, their performance often degrades under strong noise interference, which limits their applicability in real-world industrial scenarios. To address this issue, this paper proposes a novel attention-enhanced Transformer model that integrates large-kernel convolution and multiscale CNN structures for robust fault diagnosis. The proposed framework effectively combines spatiotemporal feature modeling with adaptive frequency-domain enhancement, enabling it to suppress noise and highlight informative diagnostic features. Experimental results on the Paderborn University and Case Western Reserve University datasets show that the proposed method achieves superior recognition accuracy under various signal-to-noise ratios, outperforming several state-of-the-art models. Furthermore, ablation studies and visualization analyses validate the effectiveness and soundness of the proposed architecture.