Article
Computer Science, Information Systems
Christine Dewi, Rung-Ching Chen, Xiaoyi Jiang, Hui Yu
Summary: Traffic sign detection is crucial for smart vehicles, and this study analyzes the use of Yolo V4 and Yolo V4-tiny with Spatial Pyramid Pooling for efficient object detection. The results show that Yolo V4_1 with SPP outperforms previous schemes, achieving high accuracy and performance improvements.
MULTIMEDIA TOOLS AND APPLICATIONS
(2022)
Article
Environmental Sciences
Jinrui Li, Libin Chen, Jian Shen, Xiongwu Xiao, Xiaosong Liu, Xin Sun, Xiao Wang, Deren Li
Summary: Fast and high-accuracy detection of underwater targets based on side scan sonar images has great potential for various applications. An improved neural network with spatial pyramid pooling and online dataset preprocessing based on YOLO V3 is proposed to address the limitations of low-resolution images and improve detection performance. Experimental results show that the proposed method achieves higher detection accuracy compared to other algorithms.
Article
Agriculture, Multidisciplinary
Changhong Xu, Maofang Gao, Jingwen Yan, Yunxiang Jin, Guijun Yang, Wenbin Wu
Summary: This study proposes a multi-layer pyramid crop classification network (MPNet) to solve the challenges faced in crop classification tasks based on neural networks. By using a pyramid pooling module to improve global information acquisition and an information concatenation module to retain upper features, feature loss during crop extraction is reduced. Experimental results show that the proposed model achieves the highest accuracy compared to five other deep learning models, and it also has a shorter training time and higher efficiency under the same running conditions. Overall, this study improves the efficiency and accuracy of crop classification tasks in unbalanced temporal and spatial distribution, providing a feasible solution for crop classification tasks in complex growing areas.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2023)
Article
Computer Science, Artificial Intelligence
Xuhang Lian, Yanwei Pang, Jungong Han, Jing Pan
Summary: The CHASPP module overcomes the limited sampling ranges of ASPP and the loss of local information by expanding the sampling distribution through cascaded components and hierarchical pyramid pooling structure, presenting rich local detail characteristics and effectively utilizing global contextual information.
PATTERN RECOGNITION
(2021)
Article
Computer Science, Interdisciplinary Applications
Jinke Wang, Peiqing Lv, Haiying Wang, Changfa Shi
Summary: Liver segmentation is crucial for liver cancer diagnosis and surgical planning. In this paper, a modified U-Net framework incorporating SE, ASPP, and residual learning techniques was proposed for accurate and robust liver CT segmentation, showing higher accuracy compared to other closely related models on LiTS17 and SLiver07 datasets.
COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE
(2021)
Article
Radiology, Nuclear Medicine & Medical Imaging
Zhenhong Liu, Hongfang Yuan, Huaqing Wang
Summary: The morbidity of PE is only lower than that of coronary heart disease and hypertension. Early detection, early diagnosis, and timely treatment are crucial. However, automatic segmentation of PE from CT images is challenging. In this study, a deep neural network called CAM-Wnet with CA mechanisms and PPMs was proposed to accurately segment PE. The network showed promising results in improving the segmentation effect of PE in lung CT images.
Article
Chemistry, Analytical
Shanyong Xu, Yujie Zhou, Yourui Huang, Tao Han
Summary: This study proposes a coal gangue identification method based on YOLOv4-tiny and FPGA, which achieves higher energy efficiency ratio and faster recognition speed through deployment and optimization on the FPGA platform.
Article
Agriculture, Multidisciplinary
Eko Prasetyo, Nanik Suciati, Chastine Fatichah
Summary: This research proposes a modified Yolov4-tiny architecture to improve the accuracy of fish body part detection by enhancing and balancing feature diversity and attaching an extra-branch detector. The use of bottleneck and expansion convolution helps reduce computational resource usage. Experimental results show that the proposed model outperforms other models in terms of detection accuracy and is more efficient in size and resource usage.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2022)
Article
Engineering, Biomedical
Xin-Feng Du, Jie-Sheng Wang, Wei-zhen Sun
Summary: The proposed pyramid scene parseing U-Net segmentation algorithm based on attention mechanism improves the ability to obtain global information by introducing a modified PSP-Net pyramid pooling module and attention mechanism. This enhances the detection ability of blood vessel pixels, suppresses irrelevant information interference, and outperforms current U-Net algorithm and mainstream retinal vascular segmentation algorithms, as demonstrated by significant experimental results.
PHYSICS IN MEDICINE AND BIOLOGY
(2021)
Article
Computer Science, Information Systems
Kehong Liu
Summary: This paper proposes an approach called STBi-YOLO for accurately recognizing lung nodules in CT images. By modifying the network structure and loss function of YOLO-v5, and employing feature fusion and optimized model training, STBi-YOLO achieves high accuracy and recall rates in lung nodule detection, while consuming less memory than YOLO-v5.
Article
Agriculture, Multidisciplinary
Yaohai Lin, Ruixing Cai, Peijie Lin, Shuying Cheng
Summary: This article introduces a new method for log detection based on K-median clustering and improved YOLOv4-Tiny network. The method achieves accurate detection of bundled log ends by selecting appropriate anchor box sizes and enhancing feature extraction capability, resulting in improved precision, recall, and F1 score.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2022)
Article
Computer Science, Information Systems
Chaofeng Lan, Jingjuan Jiang, Lei Zhang, Zhen Zeng
Summary: With the development and widespread application of voice interaction technology, it has become crucial to improve the accuracy of blind source separation technology. This paper proposes an improved Wave-U-Net model to enhance the separation results of vocal and accompaniment. Experimental tests demonstrate that the proposed model outperforms existing baseline models in terms of separation performance.
Article
Computer Science, Information Systems
Guanghui Xue, Sanxi Li, Peng Hou, Song Gao, Renjie Tan
Summary: This paper presents a detection algorithm for coal gangue sorting robot, which achieves faster detection speed and smaller model size through the use of lightweight networks and improvements in real-time performance. It also demonstrates good gangue detection performance.
INTERNET OF THINGS
(2023)
Article
Agriculture, Multidisciplinary
Zhenwei Yu, Yuehua Liu, Sufang Yu, Zhanhua Song, Yinfa Yan, Fade Li, Zhonghua Wang, Fuyang Tian
Summary: An efficient detection method based on the improved deep learning model FS-YOLOv4 is proposed for dairy cow teats, which significantly improves detection accuracy and speed, and enhances anti-noise ability.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2022)
Article
Public, Environmental & Occupational Health
Sanchit Vijay, Thejineaswar Guhan, Kathiravan Srinivasan, P. M. Durai Raj Vincent, Chuan-Yu Chang
Summary: Brain tumor diagnosis has been time-consuming, but automating the segmentation process can speed it up. This paper introduces SPP-U-Net, a model that replaces residual connections with a combination of Spatial Pyramid Pooling (SPP) and Attention blocks, allowing for greater context and scope in the segmentation. The proposed approach achieves comparable results to existing literature without increasing training parameters.
FRONTIERS IN PUBLIC HEALTH
(2023)
