检索结果-内蒙古大学图书馆

Deep learning for plant stress detection: A comprehensive review of technologies, challenges, and future directions

COMPUTERS AND ELECTRONICS IN AGRICULTURE 2025年 229卷

作者： Paul, Nijhum Sunil, G. C. Horvath, David Sun, Xin North Dakota State Univ Dept Agr & Biosyst Engn Fargo ND 58108 USA North Dakota State Univ Genom Phen & Bioinformat Program Fargo ND 58108 USA USDA ARS ETSARC Weed & insect Biol unit Fargo ND 58102 USA

Deep learning (DL)-based systems have emerged as powerful methods for the diagnosis and treatment of plant stress, offering high accuracy and efficiency in analyzing imagery data. This review paper aims to present a thorough overview of the state-of-the-art DL technologies for plant stress detection. For this purpose, a systematic literature review was conducted to identify relevant articles for highlighting the technologies and approaches currently employed in the development of a DL-based plant stress detection system, specifically the advancement of image-based data collection systems, image preprocessing techniques, and deep learning algorithms and their applications in plant stress classification, disease detection, and segmentation tasks. Additionally, this review emphasizes the challenges and future directions in collecting and preprocessing image data, model development, and deployment in real-world agricultural settings. Some of the key findings from this review paper are: Training data: (i) Most plant stress detection models have been trained on Red Green Blue (RGB) images;(ii) Data augmentation can increase both the quantity and variation of training data;(iii) Handling multimodal inputs (e. g., image, temperature, humidity) allows the model to leverage information from diverse sources, which can improve prediction accuracy;Model Design and Efficiency: (i) Self-supervised learning (SSL) and Few-shot learning (FSL)-based methods may be better than transfer learning (TL)-based models for classifying plant stress when the number of labeled training images are scarce;(ii) Custom designed DL architectures for a specific stress and plant type can have better performance than the state-of-the-art DL architectures in terms of efficiency, overfitting, and accuracy;(iii) The multi-task learning DL structure reuses most of the network architecture while performing multiple tasks (e.g., estimate stress type and severity) simultaneously, which makes the learning much

关键词： Plant stress Computer vision Deep learning image processing Imaging sensor Precision agriculture

来源：评论

学校读者我要写书评

暂无评论

Local Geometric Indexing of High Resolution Data for Facial Reconstruction From Sparse Markers

引用

IEEE TRANSACTIONS ON VISUALIZATION AND COMPUTER GRAPHICS 2024年第8期30卷 5289-5298页

作者： Cong, Matthew Lan, Lana Fedkiw, Ronald Ind Light & Mag San Francisco CA 94129 USA Stanford Univ Dept Comp Sci Stanford CA 94305 USA

When considering sparse motion capture marker data, one typically struggles to balance its overfitting via a high dimensional blendshape system versus underfitting caused by smoothness constraints. With the current trend towards using more and more data, our aim is not to fit the motion capture markers with a parameterized (blendshape) model or to smoothly interpolate a surface through the marker positions, but rather to find an instance in the high resolution dataset that contains local geometry to fit each marker. Just as is true for typical machine learning applications, this approach benefits from a plethora of data, and thus we also consider augmenting the dataset via specially designed physical simulations that target the high resolution dataset such that the simulation output lies on the same so-called manifold as the data targeted.

关键词： Shape Faces Geometry Surface reconstruction Cameras Point cloud compression Deformation Computer graphics image processing and computer vision interpolation

来源：评论

学校读者我要写书评

暂无评论

image processing Method Based on MATLAB in the Application of Belt Tracking with Industrial Robot 13th

Image Processing Method Based on MATLAB in the Application o...

引用

13th International Conference on Computer Engineering and Networks (CENet)

作者： Tang Youan Zhao Lingyan Chen Gandong Shandong Jiaotong Univ Sch Construct Machinery Jinan Peoples R China

ISBN: (纸本)9789819992454;9789819992430;9789819992423

At present, the application of industrial robots combined with visual systems to achieve dynamic grasping materials of the belt is becoming increasingly widespread. Unlike the behavior of industrial robots grabbing after the belt stops, industrial robots dynamically tracking and grabbing materials on the belt can greatly improve production efficiency. For vision, processing distorted material images during high-speed movement and to obtain accurate coordinate points of materials is a key task to improve the accuracy of the belt tracking applications with industrial robot. Developing image processing algorithms based on MATLAB can, on the one hand, utilize existing software and hardware interface functions to improve development efficiency;On the other hand, autonomous and controllable image processing algorithms can be developed based on application requirements to maximize system accuracy.

关键词： Belt Dynamical tracking Industrial robots image processing

来源：评论

学校读者我要写书评

暂无评论

CONTROLLABLE UNIVERSAL EDGE-PRESERVING image FILTERING 7

CONTROLLABLE UNIVERSAL EDGE-PRESERVING IMAGE FILTERING

引用

7th IEEE International Conference on Multimedia Information processing and Retrieval (MIPR)

作者： Liang, Shijun Fu, Dongdong Michigan State Univ Dept Biomed Engn E Lansing MI 48824 USA Dolby Labs Inc Sunnyvale CA USA

ISBN: (纸本)9798350351439;9798350351422

In this study, we investigate the Deep image Prior (DIP) in enhancing image smoothing, a crucial component in numerous computer vision and graphics applications. Although deep learning has demonstrated remarkable achievements in these domains, it often falls short in flexibility and controllability, in contrast to traditional methods, which are more adaptable and typically exhibit subpar performance. Notably, some end-to-end deep learning models offer control over edge preservation, yet their performance remains marginally suboptimal. To address this shortcoming, we introduce an innovative network architecture that diverges from the traditional U-Net model, featuring a Laplacian pyramid as the encoder and a deep decoder as the decoding component, integrated with a bilateral filter loss to improve DIP. This design aids the network in rapidly assimilating essential low-frequency information. Our approach excels in retaining texture details, significantly improving image smoothing and related tasks beyond the capabilities of standard DIP methods. Moreover, our technique outperforms the leading unsupervised method, pyramid texture filtering, in texture filtering tasks and other applications.

关键词： image smoothing machine learning deep learning Deep image prior

来源：评论

学校读者我要写书评

暂无评论

Revisiting Token Pruning for Object Detection and Instance Segmentation

Revisiting Token Pruning for Object Detection and Instance S...

引用

IEEE/CVF Winter Conference on applications of Computer vision (WACV)

作者： Liu, Yifei Gehrig, Mathias Messikommer, Nico Cannici, Marco Scaramuzza, Davide Univ Zurich Robot & Percept Grp Zurich Switzerland

ISBN: (纸本)9798350318920;9798350318937

vision Transformers (ViTs) have shown impressive performance in computer vision, but their high computational cost, quadratic in the number of tokens, limits their adoption in computation-constrained applications. However, this large number of tokens may not be necessary, as not all tokens are equally important. In this paper, we investigate token pruning to accelerate inference for object detection and instance segmentation, extending prior works from image classification. Through extensive experiments, we offer four insights for dense tasks: (i) tokens should not be completely pruned and discarded, but rather preserved in the feature maps for later use. (ii) reactivating previously pruned tokens can further enhance model performance. (iii) a dynamic pruning rate based on images is better than a fixed pruning rate. (iv) a lightweight, 2-layer MLP can effectively prune tokens, achieving accuracy comparable with complex gating networks with a simpler design. We assess the effects of these design decisions on the COCO dataset and introduce an approach that incorporates these findings, showing a reduction in performance decline from similar to 1.5 mAP to similar to 0.3 mAP in both boxes and masks, compared to existing token pruning methods. In relation to the dense counterpart that utilizes all tokens, our method realizes an increase in inference speed, achieving up to 34% faster performance for the entire network and 46% for the backbone. Code: https://***/uzh-rpg/svit/

关键词： Algorithms Algorithms and algorithms formulations image recognition and understanding machine learning architectures

来源：评论

学校读者我要写书评

暂无评论

A deep journey into image enhancement: A survey of current and emerging trends

引用

INFORMATION FUSION 2023年第1期93卷 36-76页

作者： Lepcha, Dawa Chyophel Goyal, Bhawna Dogra, Ayush Sharma, Kanta Prasad Gupta, Deena Nath Chandigarh Univ Dept ECE Mohali 140413 Punjab India Ronin Inst Montclair NJ 07043 USA GLA Univ Inst Engn & Technol Mathura India C DAC Mumbai Mumbai India

image captured under poor-illumination conditions often display attributes of having poor contrasts, low brightness, a narrow gray range, colour distortions and considerable interference, which seriously affect the qualitative visual effects on human eyes and severely restrict the efficiency of several machine vision systems. In addition, underwater images often suffer from colour shift and contrast degradation because of an absorption and scattering of light while travelling in water. These unpleasant effects limits visibility, reduce contrast and even generate colour casts that limits the use of underwater images and videos in marine archaeology and biology. In medical imaging applications, medical images are important tools for detecting and diagnosing several medical conditions and ailments. However, the quality of medical images can often be degraded during image acquisition due to factors such as noise interference, artefacts, and poor illumination. This may lead to the misdiagnosis of medical conditions, which can further aggravate life threatening situations. image enhancement is one of the most important technologies in the field of image processing, and its purpose is to improve the quality of images for specific applications. In general, the basic principle of image enhancement is to improve the quality and visual interpretability of an image so that it is more suitable for the specific applications and the observers. Over the last few decades, numerous image enhancement techniques have been proposed in the literature This study covers a systematic survey on existing state-of-the-art image enhancement techniques into broad classification of their algorithms. In addition, this paper summarises the datasets utilised in the literature for performing the experiments. Furthermore, an attention has been drawn towards several evaluation parameters for quantitative evaluation and compared different state-of-the-art algorithms for performance analysis on benchmark

关键词： Review image enhancement Fuzzy theory Retinex theory Deep learning Convolutional neural networks (CNNs) Generative adversarial networks (GANs) applications Quality assessment criteria Survey

来源：评论

学校读者我要写书评

暂无评论

Automatic imagery Bank Cheque data extraction based on machine learning approaches: a comprehensive survey

引用

MULTIMEDIA TOOLS AND applications 2023年第20期82卷 30543-30598页

作者： Thakur, Neha Ghai, Deepika Kumar, Sandeep Lovely Profess Univ Phagwara 144411 Punjab India Koneru Lakshmaiah Educ Fdn Vaddeswaram Andhra Pradesh India

Bank Cheques are used mainly for financial transactions due to which they are processed in enormous amounts on daily basis around the globe. Often, Cheque execution time and expenses can be saved if the whole method of recognition and verification of the Cheque becomes automatic. Automatic bank Cheque processing system is an emerging research field in the area of computer vision, image processing, pattern recognition, machine learning, and deep learning. The article emphasizes the stages of the proceedings of image acquisition, pre-processing, and extraction and recognition in the automatic bank Cheque processing system. This paper describes the various steps involved in the system of automatic data extraction. It further classifies and examines existing challenges in different stages of automated processing of bank Cheques. An attempt is made in this paper to present state-of-the-art techniques for the automatic processing of bank Cheque images. The categories and sub-categories of various fields related to bank Cheque images are illustrated, benchmark datasets are enumerated, and the performance of the most representative approaches is compared. Moreover, it also contains some information about the products available in the market for automatic Cheque processing. This review provides a fundamental comparison and analysis of the remaining problems in the field. It is found that multilayer feed-forward neural network gave an accuracy of 97.31% for payee's name recognition systems;HMM-MLP gave an accuracy of 95.5% for date recognition system. In the courtesy and legal amount system, DNN gave an accuracy of 98.5% for digit recognition, MLP gave an accuracy of 93.2% for courtesy amount, MQDF gave an accuracy of 97.04% for the legal amount. Further, the SVM classifier gave an accuracy of 99.13% for signature recognition, and deep learning-based Convolutional Neural Networks (CNN) gave an accuracy of 99.14% for handwritten numeric character recognition. This survey paper

关键词： Bank Cheque processing Courtesy amount recognition Date recognition Legal amount recognition MICR code Signature verification

来源：评论

学校读者我要写书评

暂无评论

Restoration of motion-blurred numeral image using a complex-amplitude diffractive processor

引用

OPTICS LETTERS 2024年第17期49卷 4914-4917页

作者： Zhu, Haodong Yin, Ruiqi Hu, Tie Xia, Rui Li, Minglong Zhao, Ming Yang, ZhenYu Huazhong Univ Sci & Technol Sch Opt & Elect Informat Nanophoton Lab Wuhan 430074 Peoples R China

We propose a complex-amplitude diffractive processor based on diffractive deep neural networks (D2NNs). By precisely controlling the propagation of an optical field, it can effectively remove the motion blur in numeral images and realize the restoration. Comparative analysis of phase-only, amplitude-only, and complex-amplitude diffractive processor reveals that the complex-amplitude network significantly enhances the performance of the processor and improves the peak signal-to-noise ratio (PSNR) of the images. Appropriate use of complex-amplitude networks contributes to reduce the number of network layers and alleviates alignment difficulties. Due to its fast processing speed and low power consumption, complex-amplitude diffractive processors hold potential applications in various fields including road monitoring, sports photography, satellite imaging, and medical diagnostics. (c) 2024 Optica Publishing Group. All rights, including for text and data mining (TDM), Artificial Intelligence (AI) training, and similar technologies, are reserved.

关键词： image restoration machine vision Neural networks Optical computing Optical fields Photography

来源：评论

学校读者我要写书评

暂无评论

LFRNet: Low-Light Face Super-Resolution with Light Frequency Representation 2

LFRNet: Low-Light Face Super-Resolution with Light Frequency...

引用

2nd International Conference on Algorithm, image processing and machine vision, AIPMV 2024

作者： Zha, Bingxin Zhu, Huijie Zhang, Haolin Yang, Shengying School of Information and Electronic Engineering Zhejiang University of Science and Technology Hangzhou310023 China Huzhou Zhongke Fanzai Electric Power Technology Development Co. Ltd Huzhou313000 China Co. Ltd Shanghai200233 China

ISBN: (纸本)9798350390254

In the field of computer vision, the task of facial super-resolution (FSR) is crucial for applications such as surveillance and photo restoration. However, factors such as noise and artifacts in real-world scenarios severely degrade image quality. Although existing methods use geometric priors and facial heatmap localization to improve FSR performance, these priors are often inaccurate under low-light conditions, affecting the results. This paper proposes an innovative low-light facial enhancement network aimed at integrating FSR and low-light enhancement tasks. We designed a Light Frequency Inference Block (LFIB) to capture and refine brightness and texture features at different frequencies, combining it with Transformer modules in an Encoder-Decoder architecture. The LFIB module separates degraded images into low-frequency and high-frequency brightness features and employs spatial cross-attention to capture facial texture details, effectively addressing the degradation issues in facial images. Experimental results demonstrate that our method excels in the low-light facial super-resolution task, outperforming existing methods on various metrics. Additionally, it shows good generalization capabilities in real-world scenarios, confirming its potential for practical applications. © 2024 IEEE.

关键词： image reconstruction

来源：评论

学校读者我要写书评

暂无评论

image Classifier Using Resource-Constrained Device and Tiny machine Learning 1st

Image Classifier Using Resource-Constrained Device and Tiny ...

引用

1st International Conference on Data Engineering and machine Intelligence, ICDEMI 2023

作者： Vinod, K.S. Kanmani Ruby, E.D. Vel Tech Rangarajan Dr. Sagunthala R&D Institute of Science and Technology Chennai India

ISBN: (纸本)9789819776153

image classification is one of the main parts of computer vision, which is important in applications like self-driving automotives/vehicle systems. While working with image/video data it needs huge amount of resources including computing power, graphic processing units (GPU), memory, high end CPUs, etc. We can use the small microcontrollers to do the same task, by using high-end machines for training and building the model and converting the model so that they fit into microcontroller unit, by using the method called Transfer Learning. In our work, we use Arduino nano 33 BLE sense and OV7675 camera module and online machine learning framework called Edge impulse for building the model. It is found that our tiny machine learning model works well and provides a real-time solution for image classification in the resource-constrained scenario. The experimental results show that the image classifier is performing with around 100% accuracy and so has got a wide scope in real-time classification applications. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Graphics processing unit

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：