Article
Environmental Sciences
Mohamed Marzhar Anuar, Alfian Abdul Halin, Thinagaran Perumal, Bahareh Kalantar
Summary: In recent years, complex food security issues caused by climatic changes, limitations in human labour, and increasing production costs require a strategic approach. The emergence of artificial intelligence, especially deep learning algorithms in computer vision, offers a new alternative to address these problems in the agriculture industry. This study explores different deep convolutional neural network models combined with state-of-the-art feature extractors to improve the detection of defective paddy rice seedlings. Experimental results demonstrate the effectiveness of the proposed methods in achieving high precision and F1-Score using EfficientDet-D1 EfficientNet as a one-stage pretrained object detector.
Article
Computer Science, Artificial Intelligence
Klaas Dijkstra, Jaap van de Loosdrecht, Waatze A. Atsma, Lambert R. B. Schomaker, Marco A. Wiering
Summary: CentroidNetV2 is a novel hybrid Convolutional Neural Network specifically designed to segment and count many small and connected object instances. It achieves high-quality centroids and borders of object instances by decoding centroid votes and border votes, using a loss function that combines cross-entropy loss and Euclidean-distance loss.
Article
Agriculture, Multidisciplinary
Chenjiao Tan, Changying Li, Dongjian He, Huaibo Song
Summary: This study improves the tracking method of cotton seedlings by incorporating a one-stage object detection network and optical flow technology, achieving fast and accurate counting. The proposed method shows high accuracy and speed in tracking cotton seedlings, which can be applied to plant breeding and agricultural management.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2022)
Article
Computer Science, Artificial Intelligence
Yang Liu, Peng Sun, Nickolas Wergeles, Yi Shang
Summary: This paper reviews deep learning methods for small object detection, discussing challenges, solutions, and techniques. Experimental results show that Faster R-CNN performs the best in detecting small objects.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Computer Science, Artificial Intelligence
Zelong Kong, Yongquan Chen, Xinping Guan, Xinyi Le
Summary: Object detection is a significant field in computer vision, but the imbalance problem negatively affects performance. This research reveals two sources of imbalance in existing object detection methods and proposes solutions in terms of model architecture and optimization target. By introducing a location scale equilibrium module and a repulsive loss, the proposed method can address the imbalance in the location distribution of objects with different sizes and the representation information of different categories of objects in practical applications.
Article
Chemistry, Multidisciplinary
Deema Moharram, Xuguang Yuan, Dan Li
Summary: Tree-counting methods based on computer vision technologies are low-cost and efficient alternatives to traditional methods. This study proposes a deep learning algorithm for detecting and counting tree seedlings in images, which has high economic value and broad application prospects. The method can accurately identify and count different types of tree seedlings, as demonstrated by the experimental results. The proposed method can provide technical support for tree counting tasks.
APPLIED SCIENCES-BASEL
(2023)
Article
Computer Science, Artificial Intelligence
Chengxian Li, Ming Shao, Qirui Yang, Siyu Xia
Summary: The authors propose a novel counting model to estimate the number of repetitive actions in temporal 3D skeleton data. This is the first work of its kind using skeleton data for high-precision repetitive action counting. The model follows a bottom-up pipeline to clip the sub-action and uses robust aggregation in inference. The proposed model outperforms existing video-based methods in terms of accuracy in real-time inference.
IET COMPUTER VISION
(2023)
Article
Computer Science, Artificial Intelligence
Javier Rodriguez-Vazquez, Adrian Alvarez-Fernandez, Martin Molina, Pascual Campoy
Summary: This paper presents a novel object counting method that provides accurate counts and object position information by localizing each object. The method first maps objects to blob-like structures using CNN, then gathers object positions using a LoG filter, and improves results significantly through a semi-adversarial training procedure. The method performs on par with the state of the art while offering additional position information.
Article
Computer Science, Information Systems
Tausif Diwan, G. Anirudh, Jitendra Tembhurne
Summary: Object detection is a significant problem in computer vision, and deep learning has greatly improved its performance. Object detectors can be categorized into two stage and single stage detectors, with two stage detectors typically achieving higher accuracy and single stage detectors having faster inference time. YOLO, a widely adopted single stage object detection algorithm, has the advantage of faster inference speed. This paper provides a comprehensive review of single stage object detectors, particularly YOLO, and compares them with two stage detectors. It also summarizes different versions of YOLO and their applications, as well as future research directions.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Environmental Sciences
Yuyun Pan, Nengzhi Zhu, Lu Ding, Xiuhua Li, Hui-Hwang Goh, Chao Han, Muqing Zhang
Summary: This study proposes a method based on improved Faster RCNN for automatically detecting and counting sugarcane seedlings using aerial photography. The proposed method outperforms other commonly used detection models and a seedlings de-duplication algorithm is further proposed to eliminate counting errors.
Article
Engineering, Electrical & Electronic
Xu Cheng, Zhixiang Wang, Chen Song, Zitong Yu
Summary: In this study, a hierarchical feature fusion and reconstruction method called FFR-SSD is proposed to address the challenge of detecting small and multi-scale objects. By incorporating a multi-scale visual attention model, a hierarchical feature map weighing mechanism, and an effective feature map reconstruction module, the proposed method achieves significant improvements in object detection performance.
SIGNAL IMAGE AND VIDEO PROCESSING
(2023)
Review
Engineering, Electrical & Electronic
Ravpreet Kaur, Sarbjeet Singh
Summary: In the field of computer vision, Deep Convolutional Neural Networks (DCNNs) have shown excellent performance and are applied in video processing, object detection, image segmentation, image classification, speech recognition, and natural language processing. Object detection, which is crucial and challenging, has numerous applications in security, military, transportation, and medical sciences. This review paper provides a detailed overview of object detection, including frameworks, neural networks, datasets, and evaluation metrics, and discusses the improvement in performance due to the evolution of deep learning algorithms.
DIGITAL SIGNAL PROCESSING
(2023)
Article
Agriculture, Multidisciplinary
Endai Huang, Axiu Mao, Haiming Gan, Maria Camila Ceballos, Thomas D. Parsons, Yueju Xue, Kai Liu
Summary: A two-stage center clustering network (CClusnet) was developed to improve automated piglet counting performance in the presence of occlusions. Testing results showed that CClusnet achieved a mean absolute error of 0.43 per image for piglet counting, outperforming existing methods and network architectures. This technique is non-occlusion-specific and has potential for high accuracy in animal position detection and monitoring in similar settings.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2021)
Article
Chemistry, Analytical
Jonggwan Kim, Yooil Suh, Junhee Lee, Heechan Chae, Hanse Ahn, Yongwha Chung, Daihee Park
Summary: In this study, a camera-based automatic method for counting the number of pigs passing through a counting zone on a large-scale pig farm is proposed. The method utilizes deep-learning-based video object detection and tracking techniques and is able to achieve real-time and accurate pig counting. Experimental results demonstrate that the method achieves an accuracy of 99.44% and real-time execution.
Article
Agriculture, Dairy & Animal Science
Yu Zhang, Chengjun Yu, Hui Liu, Xiaoyan Chen, Yujie Lei, Tao Pang, Jie Zhang
Summary: In this study, an integrated deep learning model for automatic detection and counting of goats based on computer vision technology was proposed. The method achieved accurate automatic counting of goats in a practical breeding environment. The model was improved using advanced strategies, resulting in a significant improvement in performance.
Article
Geochemistry & Geophysics
Lei Lei, Zhenhong Li, Jintao Wu, Chengjian Zhang, Yaohui Zhu, Riqiang Chen, Zhen Dong, Hao Yang, Guijun Yang
Summary: This study presented two methods, machine learning-based and structure-based, to extract leaf base and inclination angles of maize plants. The machine learning-based method demonstrated higher estimation accuracy compared to the structure-based method. The results showed good agreement with the ground truth, indicating the effectiveness of both methods in estimating leaf base and inclination angles of maize plants.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Geochemistry & Geophysics
Daoyong Wang, Dongyan Zhang, Guijun Yang, Bo Xu, Yaowu Luo, Xiaodong Yang
Summary: A new wheat ear counting algorithm based on computer vision, utilizing SSRNet including FCNN and RCNN, was proposed to accurately and quickly count wheat ears in field conditions. The method effectively handles small sample datasets and accurately counts wheat ears in complex backgrounds.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Remote Sensing
Zhendong Sun, Qilei Zhu, Shangqi Deng, Xu Li, Xueqian Hu, Riqiang Chen, Guowen Shao, Hao Yang, Guijun Yang
Summary: This study proposes a dynamic quadripartite pixel model (DQPM) to calculate rice residue cover (RRC) in complex paddy field scenarios. By considering soil moisture content, DQPM achieves the best robustness under various soil moisture and RRC scenarios, resulting in more accurate calculations compared to traditional static models and dynamic dimidiate pixel models.
INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION
(2022)
Article
Agriculture, Multidisciplinary
Fan Wang, Chunjiang Zhao, Hao Yang, Hongzhe Jiang, Long Li, Guijun Yang
Summary: This study explores the potential of hyperspectral imaging in assessing the quality and maturity of apples. By collecting hyperspectral images and using statistical analysis, a NDSI-SCARS-PLSR model was established, which accurately estimates the firmness, soluble solids content, and starch pattern index of apples.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2022)
Article
Agriculture, Multidisciplinary
Linchao Li, Bin Wang, Puyu Feng, De Li Liu, Qinsi He, Yajie Zhang, Yakai Wang, Siyi Li, Xiaoliang Lu, Chao Yue, Yi Li, Jianqiang He, Hao Feng, Guijun Yang, Qiang Yu
Summary: This study integrated multi-source environmental variables into random forest and support vector machine models for wheat yield prediction in China. The results showed that using remotely sensed vegetation indices improved the precision of the models, with near-infrared reflectance being slightly better than other indices. The relative importance and partial dependence analyses identified the main predictors and their relationships with wheat yield.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2022)
Article
Agriculture, Multidisciplinary
Liang Han, Guijun Yang, Xiaodong Yang, Xiaoyu Song, Bo Xu, Zhenhai Li, Jintao Wu, Hao Yang, Jianwei Wu
Summary: This study uses machine learning models based on remote sensing images to detect crop lodging. The study uses Synthetic Minority Oversampling Technique and Edited Nearest Neighbors to handle imbalanced datasets, and proposes the SMOTE-ENN-XGBoost model for identifying maize lodging at the plot scale. SHapley Additive exPlanations approach is employed to interpret the features that determine lodging classification and activity prediction. The results suggest that canopy structure, spectral, and textural features should be considered simultaneously for accurate detection of crop lodging in crop breeding programs.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2022)
Article
Agronomy
Yu Zhao, Zhenhai Li, Xuexu Hu, Guijun Yang, Bujun Wang, Dandan Duan, Yuanyuan Fu, Jian Liang, Chunjiang Zhao
Summary: Timely and accurate prediction of winter wheat grain protein content is important for achieving target protein levels. A geographically weighted regression model based on meteorological factors was used to predict winter wheat GPC at the county level. The model showed higher precision than the multiple linear regressions model.
EUROPEAN JOURNAL OF AGRONOMY
(2022)
Article
Environmental Sciences
Zhenhai Li, Yu Zhao, James Taylor, Rachel Gaulton, Xiuliang Jin, Xiaoyu Song, Zhenhong Li, Yang Meng, Pengfei Chen, Haikuan Feng, Chao Wang, Wei Guo, Xingang Xu, Liping Chen, Guijun Yang
Summary: Timely monitoring of above-ground biomss is important for crop growth and yield prediction. In this study, a new crop biomass algorithm was developed to estimate winter wheat biomass using phenological observations and remote sensing data. The algorithm showed good performance in different test sites and has the potential for biomass estimation at regional scales.
REMOTE SENSING OF ENVIRONMENT
(2022)
Article
Agronomy
Yujuan Huang, Jingcheng Zhang, Jingwen Zhang, Lin Yuan, Xianfeng Zhou, Xingang Xu, Guijun Yang
Summary: This study proposed a forecasting model for the Alternaria Leaf Spot (ALS) disease in apple based on mobile internet disease survey data and high resolution spatial-temporal meteorological data. By utilizing machine learning algorithms, the study achieved an overall accuracy of 88% and Kappa of 0.53. The results demonstrated that with the aid of mobile internet technology and properly cleaned data, it is possible to collect necessary data for disease forecasting in a short time and achieve regional-scale disease prediction.
Review
Immunology
Jing Hu, Qi Guo, Congcong Liu, Qian Yu, Yuan Ren, Yueni Wu, Qin Li, Yuezhen Li, Juntao Liu
Summary: This study analyzed the immune cells of preeclampsia (PE) patients using single-cell RNA sequencing and found excessive inflammatory state in monocytes, NK cells, and B cells, as well as lower activation of memory T cells in PE patients. These findings suggest an immune imbalance in PE and provide potential therapeutic strategies for monitoring and treating the condition.
INTERNATIONAL REVIEWS OF IMMUNOLOGY
(2022)
Article
Forestry
Jinghua Wang, Xiang Li, Guijun Yang, Fan Wang, Sen Men, Bo Xu, Ze Xu, Haibin Yang, Lei Yan
Summary: In this research, the Improved YOLOv5 model was used to identify tea buds and detect germination density based on tea trees canopy visible images. The experimental results showed that the Improved YOLOv5 model achieved higher precision and recall rates compared to the original models. This research is of great significance for the scientific planning of tea bud picking and improving the production efficiency and quality of tea production.
Article
Computer Science, Software Engineering
Jintao Wu, Jingyi Zhang, Yiwen Zhang, Yiping Wen
Summary: Mobile edge computing (MEC) is a new paradigm of distributed computing that has gained attention for its ability to expand centralized cloud computing. However, the composition of micro-services in MEC is a challenging research issue that involves resource utilization, user experience, and request latency complexity.
SOFTWARE-PRACTICE & EXPERIENCE
(2023)
Article
Computer Science, Information Systems
Yifan Zhang, Jintao Wu, Hao Yang, Chengjian Zhang, Yutao Tang
Summary: Accurately obtaining the length, quantity, and distribution of fruit branches is crucial for orchard management, disease control, and improving fruit yield and quality. Edge computing has been proposed to address the challenges of efficiency and accuracy that traditional centralized computing methods face due to the diversity of fruit tree morphological structures and complex planting environments. In this study, a hierarchical growing method (HG) is proposed for edge deployment to achieve semantic and instance segmentation of fruit tree point clouds and extract phenotypic traits at the organ scale. Experimental results demonstrate that the HG method efficiently performs instance segmentation and phenotypic trait extraction with high accuracy.
Article
Agriculture, Multidisciplinary
Jing Chen, Chunjiang Zhao, Glyn Jones, Hao Yang, Zhenhong Li, Guijun Yang, Liping Chen, Yongchang Wu
Summary: Rapid socio-economic changes in China have created opportunities for the application of precision agriculture. An experiment evaluating the economic benefits of precision seeding and land leveling methods showed that they can increase crop yield and reduce soil nitrogen concentration. Considering the long-term benefits, the economic assessment needs to accurately estimate the return on investment.
ARTIFICIAL INTELLIGENCE IN AGRICULTURE
(2022)
