Article
Computer Science, Artificial Intelligence
Khurram Hameed, Douglas Chai, Alexander Rassau
Summary: This paper proposes a score-based mask edge improvement method for MaskRCNN to segment fruit and vegetable images in a supermarket environment. The introduction of a modular score-based edge improvement head and a cosine similarity based loss function can significantly improve the segmentation results of images.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Computer Science, Information Systems
Xingjian Gu, Yongjie Zhu, Shougang Ren, Xiangbo Shu
Summary: We propose a high-quality leaf instance segmentation method called BCMask, which effectively addresses the challenges of complex boundaries and occlusions among leaves. By utilizing the Bottom-up Path Augmentation module, Bilayer Occlusion Module, and Mask Refining Module, BCMask achieves accurate leaf instance segmentation. Experimental results on a chrysanthemum seedling leaf dataset collected in natural environments, as well as two public datasets (CVPPA and Komatsuna) in laboratory environments, demonstrate that BCMask outperforms state-of-the-art methods with an average precision (AP) score of 60.42%.
MULTIMEDIA SYSTEMS
(2023)
Article
Plant Sciences
Jianqiang Lu, Ruifan Yang, Chaoran Yu, Jiahan Lin, Wadi Chen, Haiwei Wu, Xin Chen, Yubin Lan, Weixing Wang
Summary: This study proposes a citrus green fruit detection method based on improved Mask-RCNN, which can effectively improve the detection accuracy through deep learning technology and is of great significance for the intelligent production of citrus.
FRONTIERS IN PLANT SCIENCE
(2022)
Article
Engineering, Biomedical
Thomas Kurmann, Pablo Marquez-Neila, Max Allan, Sebastian Wolf, Raphael Sznitman
Summary: A novel instance segmentation method is proposed for surgical instruments, surpassing previous semantic segmentation-based methods. The method provides more informative output of instance level information and achieves precise segmentation masks. Robotic instrument priors can further enhance performance.
INTERNATIONAL JOURNAL OF COMPUTER ASSISTED RADIOLOGY AND SURGERY
(2021)
Article
Computer Science, Artificial Intelligence
Rufeng Zhang, Tao Kong, Xinlong Wang, Mingyu You
Summary: Instance segmentation is a challenging task in computer vision that requires separating each instance at the pixel level. Current dominant representation for instance masks is a low-resolution binary mask. This work proposes an effective approach to encode high-resolution structured masks into a compact representation that combines high quality and low complexity. The proposed method can be easily integrated into existing pipelines and improves the mask average precision (AP) on various datasets.
PATTERN RECOGNITION
(2022)
Article
Computer Science, Information Systems
Fangjin Xu, Jianxing Huang, Jie Wu, Longyu Jiang
Summary: This paper presents a novel method for instance segmentation of sonar images and embeds it in a deep active learning framework to address the mismatch between boxIoU and NMS score and the high annotation cost. The experimental results show significant improvements on the sonar image dataset, and the proposed method achieves competitive performance with fewer labeled samples compared to other frameworks.
Article
Computer Science, Artificial Intelligence
Kai Zhao, Xuehui Wang, Xingyu Chen, Ruixin Zhang, Wei Shen
Summary: This paper focuses on partially supervised instance segmentation and proposes a teacher-student architecture where the teacher learns general knowledge and the students delve into specific categories. Extensive experiments on the COCO dataset demonstrate the superiority of this method.
Article
Agriculture, Multidisciplinary
Dandan Wang, Dongjian He
Summary: This study developed a precise apple instance segmentation method based on an improved Mask RCNN, which achieved accurate apple segmentation under various conditions and demonstrated near real-time performance. The method outperformed other comparison methods and laid the foundation for accurate fruit detection and long-term automatic growth monitoring.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2022)
Article
Engineering, Biomedical
Fatemeh Hoorali, Hossein Khosravi, Bagher Moradi
Summary: This paper aims to develop a reliable system for the automatic diagnosis of anthrax by improving the performance of Mask-RCNN. By incorporating a U-shaped structure, an enhanced FPN structure, a hybrid weighted loss function, and a dropout layer, the proposed model outperforms state-of-the-art architectures.
BIOMEDICAL SIGNAL PROCESSING AND CONTROL
(2023)
Article
Plant Sciences
Weikuan Jia, Jinmeng Wei, Qi Zhang, Ningning Pan, Yi Niu, Xiang Yin, Yanhui Ding, Xinting Ge
Summary: This study presents a two-stage instance segmentation method based on the optimized mask RCNN for fruit and vegetable picking robots. By using a lightweight backbone network and a boundary patch refinement post-processing module, the model achieves higher accuracy and segmentation quality.
FRONTIERS IN PLANT SCIENCE
(2022)
Article
Environmental Sciences
Bo Zhong, Tengfei Wei, Xiaobo Luo, Bailin Du, Longfei Hu, Kai Ao, Aixia Yang, Junjun Wu
Summary: With the rapid development of digital intelligent agriculture, accurate extraction of field information from remote sensing imagery to guide agricultural planning has become important. In this study, we analyze the scale characteristics of agricultural fields and incorporate the multi-scale idea into a Transformer model called the Multi-Swin Mask Transformer (MSMTransformer). The experimental results using the iFLYTEK Challenge 2021 Cultivated Land Extraction competition dataset show that the MSMTransformer network achieves excellent performance, outperforming other methods in terms of COCO segmentation indexes and effectively addressing the overlapping problem in dense scenes.
Article
Plant Sciences
Li Wang, Kunming Jia, Yongmin Fu, Xiaoguang Xu, Lei Fan, Qiao Wang, Wenkui Zhu, Qunfeng Niu
Summary: This study develops a new segmentation model for tobacco shred images based on an improved Mask RCNN and proposes an algorithm for calculating the overlapped area. The experimental results show that the method achieves high segmentation accuracy and overlapped area calculation accuracy, providing a new approach for similar overlapped image segmentation tasks.
FRONTIERS IN PLANT SCIENCE
(2023)
Article
Materials Science, Multidisciplinary
H. Fang, E. Hovad, Y. Zhang, D. Juul Jensen
Summary: Segmentation of spots in diffraction images is crucial for accurate grain mapping in 3D. We applied an automatic instance segmentation deep learning network based on Mask R-CNN to find spots in LabDCT images. By combining virtual noise-free images and noise-only images, we trained the network to perform better than conventional methods, resulting in improved grain reconstruction.
MATERIALS CHARACTERIZATION
(2023)
Article
Agriculture, Multidisciplinary
Pieter M. Blok, Gert Kootstra, Hakim Elchaoui Elghor, Boubacar Diallo, Frits K. van Evert, Eldert J. van Henten
Summary: The study aimed to train a CNN with fewer annotated images while maintaining its performance. An active learning method called MaskAL was developed to automatically select hard-to-classify images for annotation and retraining. The results showed that MaskAL outperformed random sampling on a broccoli dataset with visually similar classes.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2022)
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, Information Systems
Run Tian, Guiling Sun, Xiaochao Liu, Bowen Zheng
Summary: An optimized scheme for edge detection is proposed in this study, which combines the WNNM image denoising algorithm with the Sobel edge detection algorithm to enhance anti-noise performance. Experimental results demonstrate that the algorithm can achieve better detection outcomes when processing noisy images.
Article
Computer Science, Information Systems
Bowen Zheng, Jianping Zhang, Guiling Sun, Xiangnan Ren
Summary: The study introduces a trainable deep compressed sensing model, EnGe-CSNet, by combining Convolution Generative Adversarial Networks and a Variational Autoencoder to enhance the quality of image reconstruction at high compression rates. Experimental results demonstrate that the proposed model outperforms competitive algorithms at high compression rates and exhibits robustness to pattern noise in noisy images.
Article
Chemistry, Analytical
Bowen Zheng, Jianping Zhang, Guiling Sun, Xiangnan Ren
Summary: This study introduces a FLCS model, which achieves task-driven image compressed sensing through training three learnable components. Pre-trained FLCS can improve the quality of reconstructed images and reduce the running time significantly compared to existing methods.
Article
Computer Science, Information Systems
Yawen Du, Guiling Sun, Bowen Zheng, Yunlong Qu
Summary: This article presents a gateway system design method for monitoring a breeding environment based on Bluetooth low energy consumption. By prioritizing key information, adjusting sampling periods, and utilizing feedback mechanisms, the system significantly reduces transmission delay and improves data reliability. This method effectively addresses issues with Bluetooth communication efficiency in high data concurrency scenarios, enhancing communication reliability and stability while reducing losses caused by environmental mutations.
Article
Computer Science, Information Systems
Yunlong Qu, Guiling Sun, Bowen Zheng, Wang Liu
Summary: The dairy cattle breeding environment monitoring system proposed in this paper utilizes Bluetooth and B/S architecture, enhancing system stability and human-computer interaction by introducing Bootstrap responsive layout and Echarts graphical plug-in. The system solves Bluetooth connection issues and provides a good user experience, making it suitable for monitoring dairy cow growth environment with reduced deployment and usage costs.
Article
Chemistry, Analytical
Bowen Zheng, Guiling Sun, Zhaonan Meng, Ruili Nan
Summary: This paper proposes an intelligent method for vegetable recognition and size estimation using computer vision and stereo cameras in agricultural automation. Experimental results show that the method can accurately classify and estimate the size of vegetables.
Article
Computer Science, Information Systems
Fang Zhang, Guiling Sun, Bowen Zheng, Liang Dong
Summary: This paper introduces an energy management system based on the Spring Boot framework, with three layers that collect data, process it, and provide a display interface. The system utilizes RS-485 and MODBUS protocols for data collection, splits stored data horizontally in a MySQL database for improved performance, and has been operating stably for over 600 days since deployment in a manufacturing company.
Article
Computer Science, Information Systems
Xinyu Ge, Guiling Sun, Bowen Zheng, Ruili Nan
Summary: The voice encryption device described in the paper uses a composite encryption method to encrypt speech under various analog voice call conditions, ensuring voice delay, quality, and encryption effect. Compared to traditional time-domain encryption, it effectively eliminates the remaining original voice information in the encrypted data, enhancing voice security.