Article
Chemistry, Analytical
Yihong Zhang, Hang Ge, Qin Lin, Ming Zhang, Qiantao Sun
Summary: An improved maritime object detection algorithm, SRC-YOLO, is proposed to address the issues of false detection, missed detection, and low detection accuracy in foggy environments. The algorithm applies a preprocessing algorithm, a modified module, and an attention module to improve detection performance. Experimental results show that the improved SRC-YOLO effectively detects marine targets in foggy scenes.
Article
Biochemistry & Molecular Biology
Huiqing Wang, Hong Zhao, Zhiliang Yan, Jian Zhao, Jiale Han
Summary: The paper introduces a multilane dense convolutional attention network, MDCAN-Lys, to identify succinylation sites by extracting sequence information and feature space to optimize the network's abstraction ability. The experimental results demonstrate that MDCAN-Lys can recognize more succinylation sites, providing value for the application of deep learning methods in identifying succinylation sites.
Article
Plant Sciences
Boteng Sun, Wei Zhou, Shilin Zhu, Song Huang, Xun Yu, Zhenyuan Wu, Xiaolong Lei, Dameng Yin, Haixiao Xia, Yong Chen, Fei Deng, Youfeng Tao, Hong Cheng, Xiuliang Jin, Wanjun Ren
Summary: This paper proposes a universal method for detecting different types of rice panicles, which combines an improved YOLOv4 model with UAV images. The method can accurately detect curved rice panicles that are characterized by overlapping, blocking, and dense distribution in complex field environments.
FRONTIERS IN PLANT SCIENCE
(2022)
Article
Computer Science, Artificial Intelligence
Haile Zong, Chengming Qi
Summary: We propose an improved YOLOV4-tiny traffic sign recognition algorithm for easy deployment on mobile or embedded devices to address the problems of a large number of parameters and low recognition accuracy in complex scenarios. The model utilizes Octave Convolution to reduce the redundancy of low-frequency features and employs a convolutional block attention module to strengthen the weights of traffic sign regions and reduce the weights of invalid features. Additionally, the Feature Pyramid Network structure is replaced by the Simplified Path Aggregation Network structure to enhance feature information and reduce the miss detection rate.
INTERNATIONAL JOURNAL OF PATTERN RECOGNITION AND ARTIFICIAL INTELLIGENCE
(2023)
Article
Engineering, Multidisciplinary
Shijie Xu, Rui Yuan, Yong Lv, Huangxing Hu, Tong Shen, Weihang Zhu
Summary: In this paper, an intelligent approach for rolling bearing fault diagnosis using intrinsic feature extraction and CBAM-enhanced InceptionNet is introduced. The method decomposes the original signal into multiple BLIMFs using VMD and transforms them into time-frequency images using the continuous wavelet transform. The obtained images are then fed into the CBAM-enhanced InceptionNet for fault state diagnosis. Experimental results demonstrate the stable and reliable accuracy of the method, as well as its ability to reduce network model parameters and improve diagnosis efficiency.
MEASUREMENT SCIENCE AND TECHNOLOGY
(2023)
Article
Computer Science, Hardware & Architecture
Mengjiao Li, Hao Wang, Zhibo Wan
Summary: During the production and processing of steel strips, surface defects can negatively impact their integrity and functionality. Traditional defect detection methods are insufficient, so we propose an improved YOLOv4 algorithm for steel strip surface defect detection.
COMPUTERS & ELECTRICAL ENGINEERING
(2022)
Article
Computer Science, Information Systems
Shaikh Abdus Samad, J. Gitanjali
Summary: Feature space enrichment is crucial for the development of attention mechanisms in CNNs. The research presents SCMA, an attention mechanism that combines channel and spatial attention to extract features efficiently while balancing parameter efficiency and accuracy.
Article
Environmental Sciences
Yanfen Li, Hanxiang Wang, L. Minh Dang, Hyoung-Kyu Song, Hyeonjoon Moon
Summary: Object detection on remote sensing images faces unique challenges compared to natural images, such as low resolution, complex backgrounds, and variations in scale and angle. In this study, a novel framework (ORCNN-X) was proposed for oriented small object detection in remote sensing images. The framework adopts a multiscale feature extraction network (ResNeSt+) with a dynamic attention module (DCSA) and an effective feature fusion mechanism (W-PAFPN) to enhance the model's perception ability and handle variations in scale and angle. Experimental results demonstrate the state-of-the-art performance of the proposed framework in terms of detection accuracy and speed.
Article
Environmental Sciences
Zhenfang Qu, Fuzhen Zhu, Chengxiao Qi
Summary: The paper aims to improve the performance of the YOLO algorithm in remote sensing image target detection by further enhancing the YOLOv3 model, achieving improved detection speed and accuracy. The experimental results demonstrate that the optimized network model outperforms the original YOLOv3 model with the auxiliary network, with a higher mAP and increased detection frame rate.
Article
Ecology
Xiaoxue Fu, Yong Liu, Yuhai Liu
Summary: This paper proposes the YOLOT algorithm, a quantitative detection algorithm based on the improved YOLOv4, for detecting marine benthos. By introducing the Transformer mechanism and probabilistic anchor assignment, the algorithm enhances the feature extraction capability and adaptability in complex environments. Experimental results show that YOLOT achieves higher recognition precision on the marine benthic dataset compared to the original YOLOv4.
ECOLOGICAL INFORMATICS
(2022)
Article
Multidisciplinary Sciences
Thanaporn Viriyasaranon, Jang-Hwan Choi
Summary: This paper improves the performance of object detectors by modifying the backbone architecture and feature pyramid using Neural Architecture Search (NAS) and Capsule Network. The NAS-gate convolutional module deals with object scale variation, while the Capsule Attention module optimizes feature representation and localization capability. Results show that NASGC-CapANet outperforms the baseline models on multiple datasets.
SCIENTIFIC REPORTS
(2022)
Article
Multidisciplinary Sciences
Shijia Zhao, Jiachun Zheng, Shidan Sun, Lei Zhang
Summary: This study proposes an underwater object detection algorithm based on YOLOv4-tiny, which achieves better performance with less computational cost. The algorithm improves the detection accuracy by introducing a symmetric bottleneck-type structure and a symmetric FPN-Attention module. The proposed YOLO-UOD outperforms the baseline model on the Brackish underwater dataset and shows effective detection in scenarios with limited computing power.
Article
Engineering, Multidisciplinary
Mengmeng Zhao, Zhouzhou Zheng, Yingwei Sun, Yankang Chang, Chengliang Tian, Yan Zhang
Summary: This paper proposes an efficient tire defect online detection method named MSANet based on an improved lightweight YOLOv4-tiny network. The method utilizes a novel multi-scale self-attention feature enhancement module (MSAM) and an improved feature pyramid network (MC-FPN) to extract rich context information and enhance feature representation. Experimental results demonstrate the effectiveness and efficiency of the proposed method, which can meet the requirements of industrial online detection and has good generalization ability.
MEASUREMENT SCIENCE AND TECHNOLOGY
(2022)
Article
Energy & Fuels
Bin Liu, Qingda Kong, Hongyu Zhu, Dongdong Zhang, Hui Hwang Goh, Thomas Wu
Summary: This paper proposes an IDETR deep learning target detection model based on Deformable DETR combined with transfer learning and a convolutional block attention module, which can identify foreign object shading on the surfaces of PV modules in actual operating environments. This study contributes to the optimal operation and maintenance of PV systems.
Article
Engineering, Biomedical
Zhi Liu, Sixin Luo, Yunhua Lu, Yihao Zhang, Linfeng Jiang, Hanguang Xiao
Summary: According to the World Health Organization, the prevalence of somnipathy is increasing globally. Automatic sleep staging is crucial for assessing sleep quality and diagnosing psychiatric and neurological disorders caused by somnipathy. While researchers have employed deep learning methods for sleep stage classification, the modeling of intrinsic characteristics of salient waves in different sleep stages and identification of transition rules between stages remain challenging. Furthermore, the class imbalance problem in datasets hampers robust classification models. To address these issues, a deep neural network combining MSE-based U-structure and CBAM is proposed to extract multi-scale salient waves from single-channel EEG signals. The experimental results on public datasets demonstrate the superiority of the proposed model compared to existing methods.
IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING
(2023)
