The proceedings contain 158 papers. The topics discussed include: duct inspection and monitoring robot;deep learning-based approaches for preventing and predicting wild animals disappearance: a review;classification a...
ISBN:
(纸本)9798350394528
The proceedings contain 158 papers. The topics discussed include: duct inspection and monitoring robot;deep learning-based approaches for preventing and predicting wild animals disappearance: a review;classification and tracking of items on a moving conveyor belt using convolutional networks and imageprocessing;critical analysis of the 220/110/20 kV Sardanesti power substation from Romania in the context of identification elements of instability and insecurity;machine learning based collaborative prediction of SSD failures in the cloud;the impact of explainable ai on low-accuracy models: a practical approach with movie genre prediction;utilizing transfer learning-based algorithms for breast ultrasound data in multi-instance classification;predictive maintenance model-based on multi-stage neural network systems for wind turbines;and using teaching learning-based optimization with convolutional neural network to detect pneumonia based on chest X-Ray images.
In view of the potential themes and trends of Chinese language education technology, this paper proposes a technical potential theme and trend scheme based on STM structure topic model. Firstly, the influencing factor...
In view of the potential themes and trends of Chinese language education technology, this paper proposes a technical potential theme and trend scheme based on STM structure topic model. Firstly, the influencing factors is accurately located through the gradient descent theory, and the indicators is reasonably divided to reduce the interference, and the STM structure topic model is used to construct the potential themes and trends of the technology. Experimental results show that under certain evaluation criteria, the proposed scheme is superior to the traditional image recognition algorithm in terms of the accuracy of potential themes and trends of the technology, and the processing time of influencing factors, which has obvious advantages. The potential themes and trends of technology play an extremely important role in Chinese language education, which can accurately predict and optimize the growth characteristics and product generation of Chinese language education. However, traditional image recognition algorithms have certain limitations in solving potential theme simulation problems, especially when dealing with complex problems. In this paper, this paper proposes a technical potential theme and trend scheme based on the STM structure topic model to better solve this problem. In this scheme, the influencing factors is accurately located through the gradient descent theory, so as to determine the division of indicators, and the STM structure topic model is used to construct the scheme. Experimental results show that under certain evaluation criteria, the accuracy and speed of the scheme is significantly improved for different problems, and it has better performance. Therefore, the simulation scheme based on the STM structure topic model can better solve the limitations of traditional image recognition algorithms and improve the accuracy and efficiency of simulation in the potential themes and trends of Chinese language education technology.
Medical image segmentation plays a pivotal role in computer-aided diagnosis by facilitating the extraction of essential features necessary for disease detection and treatment strategies. The continuous progress in ima...
详细信息
ISBN:
(数字)9798331506520
ISBN:
(纸本)9798331506537
Medical image segmentation plays a pivotal role in computer-aided diagnosis by facilitating the extraction of essential features necessary for disease detection and treatment strategies. The continuous progress in imageprocessing technologies has led to the development of numerous segmentation methods, encompassing traditional algorithms, machine learning (ML)-driven approaches, and cutting-edge deep learning (DL) techniques. This study undertakes a comparative evaluation of these methods, focusing on their efficiency, accuracy, and suitability across different medical imaging modalities. It also delves into prominent segmentation techniques like thresholding, region-based methods, edge detection, graph cuts, active contour models, and convolutional neural networks (CNNs). Additionally, the paper explores ongoing challenges and prospective advancements aimed at enhancing segmentation efficacy in medical imaging.
The global prevalence of visual impairment and blindness due to diabetic retinopathy can be significantly reduced through improved diabetes management. Diabetic retinopathy (DR) screening is crucial as it can prevent ...
详细信息
In this paper, artificial intelligence digital imageprocessing technology is used to process power images to form an automatic power image screening system. This method can replace the traditional artificial power im...
详细信息
image denoising is a classical but still popular research topic. Removing noise from corrupted images is an indispensable step for many practical applications. Deep Learning for image denoising has shown favorable per...
详细信息
One of the most important aspects of the agricultural economy is the production of cotton, which is threatened by diseases that lower crop quality and yield. Conventional techniques for diagnosing diseases are frequen...
详细信息
Sports image classification is a complex problem with many different sports involved. It has subpar detection performance and challenges with feature recognition. The issue of classifying 110 different sports image ca...
详细信息
In this paper, we focus on the problem of bridge crack detection. In view of the characteristics of bridge cracks, such as the existence of multiple scales and shapes, and the interference of complex backgrounds and d...
详细信息
ISBN:
(数字)9798350389579
ISBN:
(纸本)9798350389586
In this paper, we focus on the problem of bridge crack detection. In view of the characteristics of bridge cracks, such as the existence of multiple scales and shapes, and the interference of complex backgrounds and different illumination conditions on the surface of bridges, we select the bridge crack images covering different scales, types, and illumination conditions to construct a dataset, and then carry out the image enhancement processes, such as rotating, randomly erasing, and adjusting the brightness and contrast of the cracks. Three deep learning algorithms, ResNet18, MobileNetV2 and EfficientNetB0, are selected for the experiments, and the performance of the models is evaluated by comparing the F1 scores, recall rates, confusion matrices, and observed loss curves. The results show that the proposed data processing and deep learning modeling strategy is effective in bridge crack detection, and the classification effect reaches a high level, providing a feasible method for bridge crack detection.
With the continuous development of digital imageprocessingalgorithms, its application scenarios have been integrated from the simple research of a single image and a single algorithm to a multi-algorithm fusion anal...
详细信息
ISBN:
(数字)9798331509828
ISBN:
(纸本)9798331509835
With the continuous development of digital imageprocessingalgorithms, its application scenarios have been integrated from the simple research of a single image and a single algorithm to a multi-algorithm fusion analysis paradigm. Therefore, this study proposes a digital media art image analysis framework that combines multiple computer imageprocessing (CIP) algorithms. This breakthrough covers three core architectures: image edge extraction, image style transfer, and image compression. In the image edge extraction module, this study designed an improved edge detection algorithm based on de-noising auto-encoder. This algorithm improves the accuracy of edge detection through multi-directional feature extraction and (I, O)-fuzzy rough set optimization, while maintaining global stability. In the image style transfer module, this study proposed a dual-module network. The network includes a texture translation network and a perceptual loss network, and achieves cross-domain style transfer through a fusion model. At the same time, the algorithm optimizes the perceptual loss function to enhance semantic matching capabilities. In the image compression module, this study uses chaotic systems to construct an optimized measurement matrix. Specifically, through a distributed data parallel training framework, a compression method based on the LPAC gradient sparse algorithm is constructed. By testing the three modules separately, the experimental results confirm the effectiveness of the proposed algorithm.
暂无评论