With the advent of the era of big data, the imageprocessing of big data based on Computer Vision (CV) has become a research field of great concern. The purpose of this study is to explore how to improve image process...
With the advent of the era of big data, the imageprocessing of big data based on Computer Vision (CV) has become a research field of great concern. The purpose of this study is to explore how to improve imageprocessingmethods through deep learning technology to better adapt to large-scale and high-dimensional image data. We propose an innovative imageprocessing framework, based on the improved ResNet50, and introduce multi-scale feature fusion, attention mechanism and width adaptation strategies. In the task of image classification, our method has achieved remarkable improvement in accuracy, precision, recall and FI value. Through the effective fusion of multi-scale features, the model better captures the abstract information in the image and improves the classification accuracy. In the target detection task, our method shows excellent performance in mAP, false alarm rate and so on. The improved ResNet50, which introduces attention mechanism, is more reliable in identifying the target position and category. The experimental results further prove the innovation and practicability of our proposed method. Compared with the traditional model and the latest imageprocessingmethods, our method shows significant advantages in big data imageprocessing. This provides new ideas and technical support for solving large-scale and diversified image data processing problems. The contribution of this study is to provide a new solution for big data imageprocessing based on CV. The improved ResNet50 model has made remarkable progress in performance, which opens up a new research direction for deep learning research in the field of big data imageprocessing.
Hyperspectral imageprocessing plays a crucial role in various applications, including remote sensing and agricultural monitoring, yet it poses significant computational challenges due to the high dimensionality and i...
详细信息
ISBN:
(数字)9798350388602
ISBN:
(纸本)9798350388619
Hyperspectral imageprocessing plays a crucial role in various applications, including remote sensing and agricultural monitoring, yet it poses significant computational challenges due to the high dimensionality and intricate data structures involved. Traditional software-based methods often suffer from high processing time and latency, limiting their effectiveness in real-time applications. To address these challenges, we propose a hardware accelerator implemented on the Genesys 2 Kintex-7 FPGA Development Board, designed specifically for hyperspectral image enhancement. This FPGA-based solution employs parallelprocessing and optimized algorithms, including interpolation and super-resolution techniques, to significantly improve processing efficiency. We evaluate the performance of our proposed method using the Indian Pines dataset, a standard benchmark for hyperspectral imaging. Experimental results demonstrate that our FPGA implementation achieves a processing time of 50 ms, significantly faster than CPU and GPU solutions, while also reducing power consumption and improving energy efficiency. These findings underscore the potential of FPGA technology for enhancing hyperspectral imageprocessing in resource-constrained environments.
In brain tumour image segmentation, due to the problems of brain tumours of different sizes and shapes, blurred boundaries as well as unbalanced categories, this paper proposes a 3D U-Net model with deformable convolu...
详细信息
ISBN:
(数字)9798350352719
ISBN:
(纸本)9798350352726
In brain tumour image segmentation, due to the problems of brain tumours of different sizes and shapes, blurred boundaries as well as unbalanced categories, this paper proposes a 3D U-Net model with deformable convolutions for brain tumor segmentation (D3DC U-Net) to enhance the accuracy of brain tumor image segmentation. By incorporating deformable 3D convolutions into the basic framework of 3D U-Net, the model's ability to learn various sizes and shapes of brain tumors is effectively improved, thereby enhancing segmentation accuracy. The addition of convolutional attention mechanism in D3D U-Net enhances the network's feature extraction capability, strengthens spatial information acquisition, and enables more accurate identification of brain tumor regions. The use of a hybrid loss function alleviates the issue of brain tumor data imbalance. Extensive experiments on the BraTS2020 dataset demonstrate superior performance of the proposed network in brain tumor segmentation of MRI images compared to other medical image segmentation methods.
Infrared and visual image fusion aims to integrate the salient and complementary features of the infrared image and visual image into one informative image. To achieve this purpose, we have proposed an infrared and vi...
详细信息
Federated Learning (FL) has emerged as a promising approach for distributed machine learning, enabling clients to collaboratively train models without sharing their data. However, existing FL methods continue to face ...
详细信息
ISBN:
(数字)9798350368741
ISBN:
(纸本)9798350368758
Federated Learning (FL) has emerged as a promising approach for distributed machine learning, enabling clients to collaboratively train models without sharing their data. However, existing FL methods continue to face challenges when dealing with non-IID data, particularly under conditions of extreme label skew. This divergence among client models can lead to a significant degradation in the accuracy of the global model. To address this critical issue, we introduce the FedSe approach, which incorporates two novel components: homogeneous grouping and sequential training. First, clients are grouped based on the distribution of data labels to ensure that each group contains a balanced representation of all labels. Second, task models are trained sequentially within these groups while training occurs in parallel across groups. The final step involves aggregating the models through averaging. Extensive experimental results on several datasets with label skew demonstrate the effectiveness of FedSe, and its results surpass current state-of-the-art methods.
Satellite image classification for land cover involves determining high-resolution imagery to recognize and classify various types of land covers. This process assists in monitoring forest health, managing resources, ...
详细信息
ISBN:
(数字)9798350375442
ISBN:
(纸本)9798350375459
Satellite image classification for land cover involves determining high-resolution imagery to recognize and classify various types of land covers. This process assists in monitoring forest health, managing resources, and assessing biodiversity effectively. However, accurately classifying images for land cover is difficult because of diverse and complex nature of forest environments and varying image qualities. In this research, the Convolutional Neural Network with Huber Loss Function (CNN-HLF) is proposed to classify the satellite images accurately for landcover mapping in forest ecosystem. The CNN-HLF effectively manage the variability and complexity of forest environments which improves the performance of land cover. Initially, the Eurosat dataset is used to determine the CNN-HLF performance. Resizing and image sharpening is applied to adjust the size and enhance the clarity of features. DenseNet201 is employed for extracting features effectively from pre-processed images. The proposed CNN-HLF achieves a high accuracy of 99.87% using Eurosat dataset compared to existing methods like GoogleNet with pre-processing and Lightweight parallel CNN (LPCNN).
The Vlasov-Poisson systems of equations (VP) describes the evolution of a distribution of collisionless particles under the effect of a collective-field potential. VP is at the basis of the study of the gravitational ...
详细信息
ISBN:
(数字)9798331524937
ISBN:
(纸本)9798331524944
The Vlasov-Poisson systems of equations (VP) describes the evolution of a distribution of collisionless particles under the effect of a collective-field potential. VP is at the basis of the study of the gravitational instability of cosmological density perturbations in Dark-Matter (DM), but its range of application extends to other fields, such as plasma physics. In the case of Cold Dark Matter, a single velocity is associated with each fluid-element (or particle), the initial condition presents a stiff discontinuity. This creates problems such as diffusion or negative distribution function when a grid based method is used to solve VP. In this work we want to highlight this problem, focusing on the technical aspects of this phenomenon. By comparing different finite volume methods and a spectral method we observe that, while all integration schemes preserve the invariants of the system (e.g, energy), the physical observable of interest, i.e., the density, is not correctly reproduced. We thus compare the density obtained with the different Eulerian integration schemes with the result obtained from a reference N -body method. We point out that the most suitable method to solve the VP system for a self-gravitating system is a spectral method.
Infrared and visible image fusion is a fundamental task for imageprocessing to enhance the image quality. To highlight target and retain effective details, different from previous methods using integer gradients, we ...
详细信息
Massive datasets are typically distributed geographically across multiple sites, where scalability, data privacy and integrity, as well as bandwidth scarcity typically discourage uploading these data to a central serv...
详细信息
ISBN:
(纸本)9789082797091
Massive datasets are typically distributed geographically across multiple sites, where scalability, data privacy and integrity, as well as bandwidth scarcity typically discourage uploading these data to a central server. This has propelled the so-called federated learning framework where multiple workers exchange information with a server to learn a "centralized" model using data locally generated and/or stored across workers. This learning framework necessitates workers to communicate iteratively with the server. Although appealing for its scalability, one needs to carefully address the various data distribution shifts across workers, which degrades the performance of the learnt model. In this context, the distributionally robust optimization framework is considered here. The objective is to endow the trained model with robustness against adversarially manipulated input data, or, distributional uncertainties, such as mismatches between training and testing data distributions, or among datasets stored at different workers. To this aim, the data distribution is assumed unknown, and to land within a Wasserstein ball centered around the empirical data distribution. This robust learning task entails an infinite-dimensional optimization problem, which is challenging. Leveraging a strong duality result, a surrogate is obtained, for which a primal-dual algorithm is developed. Compared to classical methods, the proposed algorithm offers robustness with little computational overhead. Numerical tests using image datasets showcase the merits of the proposed algorithm under several existing adversarial attacks and distributional uncertainties.
In recent years, machine learning methods have been extensively studied in Alzheimer's disease (AD) prediction. Most existing methods extract the handcraft features from images and then train a classifier for pred...
详细信息
ISBN:
(数字)9781510650435
ISBN:
(纸本)9781510650435;9781510650428
In recent years, machine learning methods have been extensively studied in Alzheimer's disease (AD) prediction. Most existing methods extract the handcraft features from images and then train a classifier for prediction. Although it has good performance, it has some deficiencies in essence, such as relying too much on image preprocessing, easily ignoring the latent lesion features. This paper proposes a deep learning network model based on the attention mechanism to learn the latent features of PET images for AD prediction. Firstly, we design a novel backbone network based on ResNet18 to capture the potential features of the lesion and avoid the problems of gradient disappearance and gradient explosion. Secondly, we add the channel attention mechanism so that the model can learn to use global information to selectively emphasize information features and suppress low-value features, which is conducive to the extraction of semantic features. Finally, we expand the data by horizontal flipping and random flipping, which reduces the over-fitting phenomenon caused by the limited medical data set and improves the generalization ability of the model. This method is evaluated on 238 brain PET images collected in the ADNI database, and the prediction accuracy is 94.2%, which is better than most mainstream algorithms.
暂无评论