In several domains, such as remote sensing, agriculture, and environmental monitoring, hyperspectral imageprocessing is essential. In this work, the Indian Pines dataset is used to investigate hyperspectral picture c...
详细信息
The composite seeker based on multi-band imaging detection is the development focus of future precision-guided weapons. A certain seeker adopts a recognition and tracking strategy based on the fusion of long-wave infr...
详细信息
Alzheimer's disease (AD) is a neurodegenerative condition that deteriorates brain cells and impairs a patient's memory. It is progressive and incurable. Early identification can shield the patient from more br...
详细信息
ISBN:
(纸本)9798350391558;9798350379990
Alzheimer's disease (AD) is a neurodegenerative condition that deteriorates brain cells and impairs a patient's memory. It is progressive and incurable. Early identification can shield the patient from more brain cell damage and, as a result, help them avoid irreversible memory loss. The scientific community has employed a number of deep learning algorithms to automatically identify Alzheimer's patients. These comprise binary classification of patient scans into stages of AD as well as moderate cognitive impairment (MCI). Limited research has been done on the multiclass classification of Alzheimer's disease (AD) up to six distinct stages. This research proposes novel technique in Alzheimer disease detection with severity level analysis utilizing deep learning (DL) model. Input is collected as MRI brain images and processed for noise removal and smoothening. Then processed image classification and disease stage is detected using pre-trained multi-layer convolutional residual transfer Random Forest with InceptionV3 model. Experimental analysis is carried out in terms of training accuracy, mean average mean average precision, sensitivity, AUC for various MRI brain image dataset. Training accuracy attained by proposed technique is 96%, mean average precision of 93%, sensitivity of 95%, AUC of 90%.
Cross-domain few-shot classification (CDFSC) is a challenging and tough task due to the significant distribution discrepancies across different domains. To address this challenge, many approaches aim to learn transfer...
详细信息
ISBN:
(纸本)9798350344868;9798350344851
Cross-domain few-shot classification (CDFSC) is a challenging and tough task due to the significant distribution discrepancies across different domains. To address this challenge, many approaches aim to learn transferable representations. Multilayer perceptron (MLP) has shown its capability to learn transferable representations in various downstream tasks, such as unsupervised image classification and supervised concept generalization. However, its potential in the fewshot settings has yet to be comprehensively explored. In this study, we investigate the potential of MLP to assist in addressing the challenges of CDFSC. Specifically, we introduce three distinct frameworks incorporating MLP in accordance with three types of few-shot classification methods to verify the effectiveness of MLP. We reveal that MLP can significantly enhance discriminative capabilities and alleviate distribution shifts, which can be supported by our expensive experiments involving 10 baseline models and 12 benchmark datasets. Furthermore, our method even compares favorably against other state-of-the-art CDFSC algorithms.
The squint imaging scenarios introduce the additional phase, leading to distortion in interferometric inverse synthetic aperture radar (InISAR) imaging results and presenting a formidable challenge for existing algori...
详细信息
This research aims to develop an advanced material detection system for conveyor belts, utilizing state-of-the-art imageprocessing and machine learning techniques to automate the identification of various materials, ...
详细信息
This paper presents an innovative technological approach to ensure the sustainability of state-controlled nature park hunting tourism and to protect wildlife. The developed smart fire control systems utilize artificia...
详细信息
ISBN:
(纸本)9783031734199;9783031734205
This paper presents an innovative technological approach to ensure the sustainability of state-controlled nature park hunting tourism and to protect wildlife. The developed smart fire control systems utilize artificial intelligence (AI) and advanced imageprocessing techniques to prevent firing by activating the safety pin in various situations. Through an integrated camera module and a minicomputer from the Jetson family, animals can be detected in real-time, and various data analysis techniques, along with object recognition algorithms, allow for ethical intervention in many situations that are invisible to the human eye and could be exploited by hunters. Particularly in various national parks, the frequent reports of accidental or uninformed hunting of rare and endangered species, especially during their breeding seasons, underline the importance of this technology. Furthermore, hunters' preference for the most dominant animal in a herd as a trophy poses a significant threat in evolutionary terms. This technology contributes significantly to the conservation of wildlife while promoting ethical and sustainable practices in hunting tourism. The study addresses the development, testing, and implementation of this technology and adds a technological dimension to wildlife conservation strategies.
With the continuous progress of computer technology, distributed intelligent systems become more and more popular. Based on this background, this paper discusses the theory of image recognition based on distributed in...
详细信息
To enhance the channel estimation accuracy in single-carrier underwater acoustic (UWA) communication, we propose integrating data reuse (DR) and proportional updating into the recursive least squares (RLS) algorithm, ...
详细信息
Soysauce-like aroms based wine needs to be stored in a dark place during long-term storage. When recognizing images of the base wine cellar, the quality of the collected images is greatly affected by environmental lig...
详细信息
ISBN:
(纸本)9798350351040;9798350351033
Soysauce-like aroms based wine needs to be stored in a dark place during long-term storage. When recognizing images of the base wine cellar, the quality of the collected images is greatly affected by environmental light, and the imaging quality is poor and there are many noise points in low light environments. The influence of ambient light poses significant challenges to computer vision. To address the issues of poor image visibility and high interference in low illuminance environments, this article proposes an improved low-illuminance image enhancement method based on Retinex-Net network. Convert the input image from RGB domain to IISV color space for processing, introduce denoising convolutional neural network Deam-Net network into the reflection image of V component for denoising, enhance the color of the illumination image of V component through spatial attention module and channel attention module, and perform bilateral filtering and contrast stretching on the S component. Finally, fuse all components and convert to RGB for obtain the enhanced image. Prove through verification have shown that the low -illumination images enhanced by the algorithm proposed in this article have improved brightness, prominent details, minimal image distortion, and are realistic and natural. They are superior to other algorithms from subjective feelings and objective evaluation indicators.
暂无评论