Article
Agriculture, Multidisciplinary
Zhuo Zhong, Juntao Xiong, Zhenhui Zheng, Bolin Liu, Shisheng Liao, Zhaowei Huo, Zhengang Yang
Summary: The accurate identification of picking points is crucial for the intelligent operation of a litchi picking robot. This paper proposed a method of locating picking points based on detecting the litchi's main fruit bearing branch (MFBB) using the YOLACT model, achieving a high precision of 89.7% and an F1 score of 83.8%. The method significantly improved the recall rate of MFBB and provided technical support for the visual recognition of the litchi picking robot.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2021)
Article
Computer Science, Artificial Intelligence
Guangyuan Yang, Rong Li, Shujin Zhang, Yuchen Wen, Xingshi Xu, Huaibo Song
Summary: This paper proposes a novel pixel-level segmentation method for cow point clouds by fusing high-resolution real images and point cloud data. The method generates 3D point clouds of cows from multi-view images using the Structure from Motion (SfM) algorithm and establishes the correspondence between point clouds and real images. It improves the segmentation accuracy of cow body regions using the improved single-stage instance segmentation algorithm YOLACT++. The experimental results demonstrate the effectiveness of the proposed method in 3D point cloud segmentation of cows.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Engineering, Marine
Jinyan Cai, Shifeng Ding, Qin Zhang, Renwei Liu, Dinghan Zeng, Li Zhou
Summary: The identification and digitization of floating ice are crucial for developing numerical ice load models for ships and marine structures in managed ice fields. This paper presents a deep learning approach for identifying broken ice blocks from images and estimating their circumferential crack size. The numerical simulation demonstrates a high accuracy in estimating the radius and open-angle of the cracks using this method.
Article
Chemistry, Multidisciplinary
Yanjie Zhu, Weidong Xu, C. S. Cai, Wen Xiong
Summary: A lightweight procedure for bridge apparent-defect detection is proposed, including crack annotation method and crack detection. The effectiveness of the proposed method is evaluated using trained models based on classic Mask RCNN and Yolact. Results show that the crack instance segmentation model can achieve a level of 90% for both accuracy and recall values with a limited dataset.
APPLIED SCIENCES-BASEL
(2023)
Article
Computer Science, Artificial Intelligence
Daniel Bolya, Chong Zhou, Fanyi Xiao, Yong Jae Lee
Summary: This paper presents a simple fully-convolutional model for real-time instance segmentation. By breaking instance segmentation into two parallel subtasks and linearly combining prototypes with mask coefficients, the model achieves competitive results with significantly faster speed. The authors also propose a faster replacement for standard non-maximum suppression and apply deformable convolutions to improve performance and efficiency.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2022)
Article
Agriculture, Multidisciplinary
Mar Ariza-Sentis, Hilmy Baja, Sergio Velez, Joao Valente
Summary: Grapevine phenotyping is the process of determining the physical properties of grape bunches and berries, and it provides valuable information for monitoring vineyard health. This study uses Multi-object tracking and segmentation (MOTS) with UAV-mounted RGB cameras to automatically extract grape phenotyping traits. The results show that the selected algorithms can accurately detect and track grape bunches, and accurately assess the number of berries per bunch.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2023)
Article
Chemistry, Multidisciplinary
Hyobin Sunwoo, Wonjun Choi, Seunguk Na, Cheekyeong Kim, Seokjae Heo
Summary: As reconstruction and redevelopment progress, the volume of construction waste increases, leading to the development of construction waste treatment technologies that utilize artificial intelligence. This study analyzes the performance differences of a construction waste recognition model after data pre-processing and labeling by individuals with varying levels of expertise, with the aim of accurately distinguishing construction waste and increasing recycling rates.
APPLIED SCIENCES-BASEL
(2022)
Article
Computer Science, Artificial Intelligence
Xia Hua, Xinqing Wang, Ting Rui, Faming Shao, Dong Wang
Summary: This paper proposes a cascaded panoptic segmentation network to address the issues existing in remote sensing image segmentation with deep convolutional neural networks. Experimental results demonstrate the effectiveness of the method by designing a shared feature pyramid network backbone, a new hybrid task cascade framework, and strategies such as learning mask quality.
APPLIED SOFT COMPUTING
(2021)
Article
Automation & Control Systems
Dehua Wei, Xiukun Wei, Qingfeng Tang, Limin Jia, Xinqiang Yin, Yang Ji
Summary: This paper investigates an innovative and intelligent method for multi-component identification and common defect detection of railway track line based on instance segmentation. A railway track line image dataset is constructed and annotated manually, and a railway track line image segmentation model (RTLSeg) is proposed. Experimental results show that the proposed method is effective and outperforms the compared baseline models.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2023)
Article
Computer Science, Software Engineering
Zichen Zhao, Xiliang Tong, Ying Sun, Dongxu Bai, Xin Liu, Guojun Zhao, Hanwen Fan, Jun Li, Cejing Zou, Baojia Chen
Summary: This article proposes an improved instance segmentation method that improves the efficiency of processing large-scale images while ensuring accuracy, and its effectiveness is demonstrated in experiments.
CONCURRENCY AND COMPUTATION-PRACTICE & EXPERIENCE
(2022)
Article
Engineering, Multidisciplinary
KeYan Ren, HaoChen Hou, SiYang Li, TianYi Yue
Summary: Lane and its bifurcation detection is a crucial topic in low cost camera-based autonomous driving and ADAS. By relabeling dataset and applying data balance strategy, a competitive lane detection model LaneDraw is developed, which improves detection speed and accuracy.
SCIENCE CHINA-TECHNOLOGICAL SCIENCES
(2021)
Article
Agronomy
Mingfeng Huang, Guoqin Xu, Junyu Li, Jianping Huang
Summary: A novel image segmentation method based on YOLACT++ with an attention module was proposed for accurately segmenting disease lesions on maize leaves. The experimental results demonstrated higher segmentation accuracy compared to existing methods under natural conditions.
Article
Computer Science, Artificial Intelligence
Jiaqi Wang, Kai Chen, Rui Xu, Ziwei Liu, Chen Change Loy, Dahua Lin
Summary: CARAFE++ is a universal, lightweight, and highly effective operator for feature reassembly in convolutional networks. It aggregates contextual information within a large receptive field, generates adaptive kernels for instance-specific content-aware handling, and introduces little computational overhead. It consistently shows significant improvements in various tasks, making it a strong building block for modern deep networks.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2022)
Article
Engineering, Marine
Brendan Chongzhi Corrigan, Zhi Yung Tay, Dimitrios Konovessis
Summary: Thousands of tonnes of litter enter the ocean every day, posing a significant threat to marine life and ecosystems. This paper focuses on the detection of ocean plastic using neural network models.
JOURNAL OF MARINE SCIENCE AND ENGINEERING
(2023)
Article
Computer Science, Artificial Intelligence
Wenchao Zhang, Chong Fu, Lin Cao, Chiu-Wing Sham
Summary: With the idea of divide and rule, there are two different forms of semantic features in two-stage instance segmentation paradigms: global features at the image level and instance features at the region-wise. The main distinction between these two macro-semantic morphological features lies in the relevance of neighborhood features caused by background noise. Therefore, we propose a more efficient methodology using Group-Inception and Asymmetric-Inception modules to enhance the representation capability of the network.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)