Article
Agriculture, Multidisciplinary
Arturo Aquino, Juan Manuel Ponce, Miguel Noguera, Jose Manuel Andujar
Summary: This paper presents a novel methodology using computer vision techniques to estimate olive fruit yield by analyzing visual features. A neural network was designed and trained with 16 descriptors, including the number and area of visible fruits. The methodology was validated on a dataset and achieved a root-mean-square-error of 0.9914 kg per sample point, with an overestimation of 2.64%.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2023)
Article
Food Science & Technology
Alen Alempijevic, Teresa Vidal-Calleja, Raphael Falque, Phillip Quin, Edwina Toohey, Brad Walmsley, Malcolm McPhee
Summary: This study introduces a technology based on 3D imaging for estimating Lean Meat Yield (LMY) of beef carcasses, showing it to be a viable and relatively accurate method.
Article
Robotics
Ji-il Park, Yeongseok Lee, Eungyo Suh, Hyunyong Jeon, Kuk-Jin Yoon, Kyung-soo Kim
Summary: This study improved the accuracy of optical flow estimation by optimizing fundamental parameters such as the number of pyramids, filters, window size, and GNC step number, instead of using deep learning methods.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2021)
Article
Robotics
Simone Andolfo, Flavio Petricca, Antonio Genova
Summary: Visual Odometry is a fundamental technique used to enhance the navigation capabilities of planetary exploration rovers. By processing the images acquired during motion, it provides estimates of the relative position and attitude between navigation steps. In this study, an independent analysis of high-resolution stereo images of the NASA Mars 2020 Perseverance rover was conducted to accurately determine its localization. The results show that the reconstructed path is consistent with the telemetered trajectory.
JOURNAL OF FIELD ROBOTICS
(2023)
Review
Agronomy
Lucas Mohimont, Francois Alin, Marine Rondeau, Nathalie Gaveau, Luiz Angelo Steffenel
Summary: This article summarizes the research on computer vision and viticulture, focusing on the use of RGB images for yield estimation and the application of artificial intelligence in grape cultivation. It also introduces methods for detecting grapevine flowers, grapes, and berries, and discusses the challenges in yield estimation.
Article
Chemistry, Analytical
Heilym Ramirez, Sergio A. Velastin, Sara Cuellar, Ernesto Fabregas, Gonzalo Farias
Summary: Recently, there has been a significant focus in the scientific community on recognizing human activity, particularly in the context of health and elderly care. While wrist-worn devices and neck pendants are already being used for activity recognition, they have limitations such as errors, discomfort, and the inability to detect subtle conditions. To overcome these challenges, image and video-based methods have been proposed, but they require advanced machine learning and deep learning techniques due to the increased complexity. This paper presents a deep learning approach with attention for activity recognition using multiple frames.
Article
Computer Science, Artificial Intelligence
Yadi Wang, Hongyun Zhang, Duoqian Miao, Witold Pedrycz
Summary: Visible-infrared person re-identification (VI-ReID) is a challenging task that addresses the limitations of conventional re-identification under insufficient illumination. It involves difficulties in pedestrian posture, camera shooting angle, background change, and cross-modality gap. This study proposes the Q-center Multi-granularity K-reciprocal Re-ranking Algorithm (QCMR) to effectively utilize both coarse-grained and fine-grained features for a comprehensive representation. Experimental results on two widely used VI-ReID benchmarks demonstrate the superiority of the proposed method, achieving state-of-the-art results with a significant improvement in mAP.
CAAI TRANSACTIONS ON INTELLIGENCE TECHNOLOGY
(2023)
Article
Robotics
Yawen Lu, Garrett Milliron, John Slagter, Guoyu Lu
Summary: In this study, a framework for simultaneous depth estimation from a single image and image focal stacks using depth-from-defocus and depth-from-focus algorithms is proposed. The system has been validated on synthetic and real datasets, surpassing existing supervised depth estimation methods by over 4% in accuracy and achieving superb performance.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2021)
Article
Computer Science, Software Engineering
S. Jeba Berlin, Mala John
Summary: In this paper, an efficient technique for human action recognition in automated video surveillance systems is proposed. The technique utilizes optical flow features and joint entropy to model human actions, and incorporates a spiking neural network to aggregate information across frames. Experimental results demonstrate the effectiveness of the proposed method.
Article
Horticulture
Marcelo H. H. Amaral, Kerry B. B. Walsh
Summary: The mass of mango fruit can be estimated by measuring the length, width, and thickness of the fruit. Prediction of fruit size and harvest time can be made based on these measurements.
Article
Computer Science, Artificial Intelligence
Nuno Pereira, Luis A. Alexandre
Summary: This article introduces a new method for estimating the six-degree-of-freedom pose of objects, which improves upon current methods in terms of accuracy and real-time usability. The method utilizes RGB-D data to segment objects and estimate their pose, employing a neural network with multiple heads to identify objects in the scene, generate masks, and estimate translation and quaternion values for rotation. The method leverages a pyramid architecture for feature extraction and fusion. Evaluation on common datasets and comparison against state-of-the-art approaches demonstrate the capabilities of MPF6D. With its low inference time and high accuracy, our method is suitable for real-time applications.
PATTERN ANALYSIS AND APPLICATIONS
(2023)
Article
Construction & Building Technology
Tong Niu, Linbo Qing, Longmei Han, Ying Long, Jingxuan Hou, Lindong Li, Wang Tang, Qizhi Teng
Summary: Small public spaces are crucial for citizens' living and socializing. This study proposes a systematic framework for quantifying vitality in small public spaces using fine-grained human trajectory data extracted from videos. A multi-indicator quantification method is utilized to comprehensively represent human vitality, resulting in a more precise evaluation model.
BUILDING AND ENVIRONMENT
(2022)
Article
Computer Science, Information Systems
Ming Zhu, Guohui Li, Qin Huang
Summary: Extracting time domain features of facial motion information can help recognize unsafe driving behaviors and reduce traffic accidents. However, it is uncertain whether facial motion can recognize unsafe behaviors of workers, and further research is needed to determine if introducing frequency domain features improves recognition accuracy.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Computer Science, Information Systems
Xiaoye Li, Bin-Bin Zhang
Summary: Vision Transformer (ViT) has shown remarkable performance in computer vision tasks, but its application in finger vein recognition is still limited. In this study, we propose a model called FV-ViT, which achieves outstanding performance in finger vein recognition by adding stringent regularization in the MLP head (regMLP) instead of modifying the ViT backbone architecture, and it outperforms other state-of-the-art methods.
Review
Health Care Sciences & Services
Lameck Mbangula Amugongo, Alexander Kriebitz, Auxane Boch, Christoph Luetge
Summary: The increasing awareness of the impact of diet on lifestyle and health has led to a rise in the use of embedded food analysis and recognition systems. These systems aim to monitor daily food consumption and provide dietary recommendations through mobile applications. However, the majority of these applications do not differentiate between food and non-food items, and they lack explanations for their classification methods. Mobile computer vision-based applications have the potential to assist in managing chronic illnesses, but they need to provide explanations to improve trust.
Article
Agriculture, Multidisciplinary
Wan-Soo Kim, Dae-Hyun Lee, Yong-Joo Kim, Taehyeong Kim, Won-Suk Lee, Chang-Hyun Choi
Summary: The study developed a machine-vision-based height measurement system for an autonomous cultivation robot, using a simple stereo camera configuration to acquire accurate height measurements of various field crops. The system showed strong agreement with manual measurements and is suitable for agricultural robot applications.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2021)
Article
Agriculture, Dairy & Animal Science
Daniel Dooyum Uyeh, Juntae Kim, Santosh Lohumi, Tusan Park, Byoung-Kwan Cho, Seungmin Woo, Won Suk Lee, Yushin Ha
Summary: A rapid and non-destructive global hyperspectral model was developed to quantify moisture content in feed materials, aiming to prevent feed spoilage and adverse health effects.
Article
Agronomy
Yangjie Shi, Xiaobo Xi, Hao Gan, Xiang Shan, Yifu Zhang, Hui Shen, Ruihong Zhang
Summary: This study introduces a row-controlled fertilizing-weeding machine to improve the accuracy of fertilizing and weeding operations and reduce environmental pollution in rice cultivation. Experimental results show that the system can achieve precise fertilization and satisfactory weeding performance at low operating speeds.
Article
Agriculture, Multidisciplinary
Changho Yun, Hak-Jin Kim, Chan-Woo Jeon, Minseok Gang, Won Suk Lee, Jong Gyu Han
Summary: This study introduces a stereovision-based auto-guidance method for tracking inter-rows between ridges and furrows, showing high accuracy above 90% in outdoor conditions. The developed algorithm effectively compensates for dynamic pitch and roll movements and robustly extracts guidance lines, making it suitable for real-world agricultural applications.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2021)
Article
Agronomy
Chuanqi Xie, Won Suk Lee
Summary: This study utilized spectral reflectance to detect citrus black spot disease, achieving accurate classification of different symptoms by collecting hyperspectral images and selecting specific wavelengths for analysis. The results demonstrated the potential of spectral reflectance information in classifying CBS symptoms effectively.
POSTHARVEST BIOLOGY AND TECHNOLOGY
(2021)
Article
Agriculture, Multidisciplinary
Amin Nasiri, Jonathan Yoder, Yang Zhao, Shawn Hawkins, Maria Prado, Hao Gan
Summary: In this study, a pose estimation-based model was developed to identify lameness in broilers for the first time. The model utilized a deep convolutional neural network and Long Short-Term Memory (LSTM) model to classify broilers based on video footages. The results showed that the model achieved high classification accuracy and could serve as an automatic and non-invasive tool for lameness assessment in poultry farms.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2022)
Article
Agriculture, Multidisciplinary
Xue Zhou, Yiannis Ampatzidis, Won Suk Lee, Congliang Zhou, Shinsuke Agehara, John K. Schueller
Summary: Bruising is a major defect in strawberries, affecting their quality and shelf life. A novel machine vision technique using deep learning was developed in this study to accurately detect and classify the severity of strawberry bruises. The technique showed higher accuracy in detection and classification under UV light.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2022)
Article
Agriculture, Dairy & Animal Science
X. Yang, Y. Zhao, H. Gan, S. Hawkins, L. Eckelkamp, M. Prado, R. Burns, J. Purswell, T. Tabler
Summary: Lameness in broilers is a significant concern for broiler production and welfare due to potential pain. Manual gait score assessment is laborious, and automatic methods are urgently needed. This study aimed to establish prediction models for automatic gait score estimations using several production and behavioral metrics. The results showed that bird weight, age, activity index, and distribution index could accurately predict broiler gait score, facilitating the development of automated assessment systems.
Article
Chemistry, Analytical
Arth M. Patel, Won Suk Lee, Natalia A. Peres
Summary: The research aimed to develop and test a more accurate leaf wetness detection system using artificial intelligence and imaging technology, which showed high accuracy in detecting wetness on a reference surface. This system can provide accurate disease risk assessment and fungicide recommendations for strawberry production.
Article
Robotics
Chapel Reid Rice, Spencer Thomas McDonald, Yang Shi, Hao Gan, Won Suk Lee, Yang Chen, Zhenbo Wang
Summary: This paper presents a drone-enabled autonomous pollination system (APS) that utilizes modules such as environment sensing, flower perception, path planning, flight control, and pollination mechanisms. The authors focus on approaches to flower perception, path planning, and flight control modules. The proposed system is tested within a model predictive control (MPC) framework, demonstrating computational savings and embedded adjustments to uncertainty.
Article
Agriculture, Multidisciplinary
Xiaobo Xi, Runyin Wang, Xintong Wang, Yangjie Shi, Ying Zhao, Baofeng Zhang, Jiwei Qu, Hao Gan, Ruihong Zhang
Summary: Improving the uniformity of multi-fertilizer mixing can enhance the overall utilization efficiency of fertilizers. This study presents a method of multi-fertilizer mixing using air blowing and blade stirring, which is optimized based on the indicator of fertilizer granules mixing uniformity. The simulation and experimental results demonstrate that the proposed method achieves better mixing uniformity compared to other mixing methods.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2023)
Article
Agriculture, Dairy & Animal Science
Mustafa Jaihuni, Hao Gan, Tom Tabler, Maria Prado, Hairong Qi, Yang Zhao
Summary: The researchers used a combination of artificial intelligence methods and computer algorithms to track individual chickens, which provided more accurate measurements of their mobility compared to traditional methods. This combined model could provide real-time and accurate information.
Article
Agriculture, Dairy & Animal Science
Amin Nasiri, Ahmad Amirivojdan, Yang Zhao, Hao Gan
Summary: Using automated approaches to investigate feeding behavior in broilers allows for accurate and non-invasive data collection, real-time monitoring, and advanced data analysis. This study aimed to estimate individual broilers' feeding time through an automated approach. The developed algorithm achieved an overall accuracy of 87.3% in estimating the feeding time.
Proceedings Paper
Agricultural Engineering
P. Puranik, W. S. Lee, N. Peres, F. Wu, A. Abd-Elrahman, S. Agehara
Summary: Production of strawberries varies weekly due to factors like flower quantity and humidity. Deep learning models and centroid detection algorithm accurately detect mature strawberries and flowers, aiding farmers in marketing and estimating harvest labor.
PRECISION AGRICULTURE'21
(2021)
Article
Agricultural Engineering
Yifan Bai, Junzhen Yu, Shuqin Yang, Jifeng Ning
Summary: A real-time recognition algorithm (Improved YOLO) is proposed in this paper for accurately identifying small, similar-colored, and overlapping strawberry seedling flowers and fruits. The experimental results show that the algorithm achieves high precision, recall, and average precision, and meets the real-time detection requirements, providing effective support for the automated management of strawberry seedling flower and fruit thinning.
BIOSYSTEMS ENGINEERING
(2024)