Today, for many applications, it is increasingly necessary to secure multimedia data. The use of full encryption has proved effective and secure, but it does not always allow a tradeoff between security and other requ...
详细信息
ISBN:
(纸本)9798331541859;9798331541842
Today, for many applications, it is increasingly necessary to secure multimedia data. The use of full encryption has proved effective and secure, but it does not always allow a tradeoff between security and other requirements. image obscuration then presents itself as a possible and welcome solution, preserving the integrity of the multimedia data while obscuring its content. In 2021, Aprilpyone and Kiya [1] proposed a block-wise bit flipping image obscuration method, using a secret key which permits the original image to be obscured and secret, while being reversible in order to recover it. However, this method presents an exploitable weakness which can be profitable for an attack. In this paper, we present an approach to attack obscured images by bit flipping, in order to reconstruct the original image without any knowledge of the secret key.
Traffic sign recognition is crucial for the safe and efficient operation of autonomous vehicles. While previous research has primarily focused on traffic sign recognition in foreign countries, these studies often face...
详细信息
ISBN:
(纸本)9798350352368
Traffic sign recognition is crucial for the safe and efficient operation of autonomous vehicles. While previous research has primarily focused on traffic sign recognition in foreign countries, these studies often face limitations such as differing traffic sign designs, language barriers in textual information, and varying environmental conditions. In this paper, we propose a traffic sign detection and recognition system tailored for Malaysia, utilizing Convolutional Neural Networks (CNNs) and Optical Character Recognition (OCR). In this paper, we propose a traffic sign detection and recognition system utilizing You Only Look Once (YOLO) V8 for object detection and EasyOCR to process textual information on selected traffic signs. Our system achieves a mean Average Precision (mAP) of 0.824 and an average processing time of 1.2 seconds per frame, which is comparable to existing literature. Furthermore, the complexity of our method is significantly reduced, enhancing its potential for real-time processingapplications, as evidenced by its efficient processing time.
This paper explores the application of the Continuous Action Learning Automata (CALA) game optimizer to Convolutional Neural Networks (CNNs) for image classification tasks. The CALA game optimizer, initially developed...
详细信息
ISBN:
(纸本)9798350384826;9798350384819
This paper explores the application of the Continuous Action Learning Automata (CALA) game optimizer to Convolutional Neural Networks (CNNs) for image classification tasks. The CALA game optimizer, initially developed for training Artificial Neural Networks (ANNs), offers a non-gradient descent-based optimization approach that can adapt to different network architectures and activation functions. Leveraging the versatility of the CALA game optimizer, we investigate its performance on CNNs, specifically targeting image recognition within the MNIST dataset. The paper discusses the rationale behind using the CALA game optimizer for CNNs, including its ability to accommodate various activation functions and deeper network architectures. Experimental results demonstrate the efficacy of CALA in training CNNs, showcasing its flexibility and effectiveness in optimizing network parameters for image classification tasks.
Cellular automata have ideal properties for scalable and efficient computing, but a lack of a training method limits their real-world applications. First, we propose to partition the cellular lattice into three region...
详细信息
ISBN:
(纸本)9798350359329;9798350359312
Cellular automata have ideal properties for scalable and efficient computing, but a lack of a training method limits their real-world applications. First, we propose to partition the cellular lattice into three regions: input, output, and processing. Second, we propose a novel synthesis method to train a linear hybrid cellular automaton. third, we show image classification on the MNIST dataset using only logic operations. By mapping local states over the globally linear lattice, the proposed model achieved above 90% test accuracy in binary image classification. Our method does not require any pre or post-processors to perform computation over the lattice. Hence, the lattice maintains its massive parallelism and locality of computation, ideal for ultra-low power processing in machine learning.
In the current landscape of rapidly expanding and diverse cloud deployments, various cloud applications are experiencing significant growth. The increasing demands from users, complexity in application development and...
详细信息
ISBN:
(纸本)9798350350920
In the current landscape of rapidly expanding and diverse cloud deployments, various cloud applications are experiencing significant growth. The increasing demands from users, complexity in application development and integration, and the lack of standardized auditing procedures have inevitably highlighted the trust issues encountered by cloud applications. Achieving trusted monitoring for cloud applications has become a challenging problem. First, different runtime states of applications exhibit distinct functional characteristics, making it difficult to establish uniform trust criteria. Second, modeling complex applications often requires substantial domain-specific knowledge, which is challenging to acquire. third, efficient and real-time trusted monitoring for cloud applications demands strong real-time performance. In this study, we propose a novel trusted monitoring method based on canonical correlation analysis (CCA). Our approach begins by modeling application states using the finite state machine theory to analyze their trustworthiness. We utilize CCA to establish correlations between application states and key system metrics, enabling runtime analysis of the trustworthiness of cloud applications by monitoring changes in these correlations. To validate and evaluate the efficiency of our proposed method, we conduct simulations of several typical untrusted scenarios of cloud applications. The results demonstrate the efficient implementation of trusted monitoring for cloud applications, enabling real-time assessment and analysis of untrustworthiness.
Dental image analysis is one of the main methods for diagnosing oral diseases. By processing dental images through artificial intelligence algorithms, the proposed diagnostic opinions can greatly improve diagnostic ef...
详细信息
ISBN:
(纸本)9798350350920
Dental image analysis is one of the main methods for diagnosing oral diseases. By processing dental images through artificial intelligence algorithms, the proposed diagnostic opinions can greatly improve diagnostic efficiency and accuracy, and have considerable medical value and potential for improvement. In recent years, the Vision Transformer model has been applied to medical image classification, and its performance has significantly outperformed CNN. This article focuses on the classification of jaw cystic lesions in oral CBCT images and classifies them based on the Vision Transformer model. The classification accuracy based on the Vision Transformer model reached 96.9%, significantly superior to other comparison models.
In the field of few-shot learning (FSL), extensive research has focused on improving network structures and training strategies. However, the role of data processing modules has not been fully explored. Therefore, in ...
详细信息
ISBN:
(纸本)9798350349405;9798350349399
In the field of few-shot learning (FSL), extensive research has focused on improving network structures and training strategies. However, the role of data processing modules has not been fully explored. Therefore, in this paper, we propose Meta-DM, a generalized data processing module for FSL problems based on diffusion models. Meta-DM is a simple yet effective module that can be easily integrated with existing FSL methods, leading to significant performance improvements in both supervised and unsupervised settings. We provide a theoretical analysis of Meta-DM and evaluate its performance on several algorithms. Our experiments show that combining Meta-DM with certain methods achieves state-of-the-art results.
Human pose estimation faces hurdles in real-world applications due to factors like lighting changes, occlusions, and cluttered environments. We introduce a unique RGB-Thermal Nearly Paired and Annotated 2D Pose Datase...
详细信息
ISBN:
(纸本)9798350349405;9798350349399
Human pose estimation faces hurdles in real-world applications due to factors like lighting changes, occlusions, and cluttered environments. We introduce a unique RGB-Thermal Nearly Paired and Annotated 2D Pose Dataset, comprising over 2,400 high-quality LWIR (thermal) images. Each image is meticulously annotated with 2D human poses, offering a valuable resource for researchers and practitioners. This dataset, captured from seven actors performing diverse everyday activities like sitting, eating, and walking, facilitates pose estimation on occlusion and other challenging scenarios. We benchmark state-of-the-art pose estimation methods on the dataset to showcase its potential, establishing a strong baseline for future research. Our results demonstrate the dataset's effectiveness in promoting advancements in pose estimation for various applications, including surveillance, healthcare, and sports analytics. The dataset and code are available at https://***/avinres/LWIRPOSE
Hyperspectral imaging has become crucial in various domains, especially for the accurate detection of human veins in medical diagnostics, though managing the extensive data from hyperspectral (HS) images remains a cha...
详细信息
ISBN:
(纸本)9798331541859;9798331541842
Hyperspectral imaging has become crucial in various domains, especially for the accurate detection of human veins in medical diagnostics, though managing the extensive data from hyperspectral (HS) images remains a challenge. To improve data handling during analysis, dimensionality reduction methods are frequently utilized. This paper presents a dimensionality reduction method for HS images using HS image inter-band cross-correlation and the K-means clustering algorithm. The proposed method computes inter-band correlations across all bands of the input HS image, which form a 2D correlation matrix. Eigen-decomposition is applied to the resulting matrix, extracting its eigenvectors and eigenvalues. The k-mean clustering algorithm is then applied to a selection of eigenvectors representing the largest eigenvalues, splitting the eigenvectors into several clusters. The reduced HS image is generated by averaging each cluster's image bands. The proposed dimensionality reduction method together with the Support Vector Machine (SVM) classifier was then used for vein detection in HS images. The HyperVein image dataset was used to generate experimental results. Experimental results were generated for the proposed method and Principal Component Analysis (PCA) and Folded PCA (FPCA). Results show the proposed method outperforms PCA and FPCA in most performance metrics.
This research explores an affordable and highprecision crowd-monitoring system of integrating data from 2D LiDAR and images from the camera through LiDAR scan data and image fusion. The novelty of the research is to a...
详细信息
ISBN:
(纸本)9798350352368
This research explores an affordable and highprecision crowd-monitoring system of integrating data from 2D LiDAR and images from the camera through LiDAR scan data and image fusion. The novelty of the research is to achieve 3-D scanning by using 2D LiDAR, which is controlled by a servo-controlled tilting mechanism to obtain multiple scan data from different elevation angles, for simulating 3D scanning operation through overlapping of multiple scanning results according to its elevation angles and performs image, 2D point cloud data fusion for human detection and distance measurement for crowd monitoring purposes. The proposed techniques enhance 2D LiDAR detection, enabling detailed scanning at lower cost and complexity. The system combines LiDAR measurements with camera imagery through proposed filtering and fusion algorithms which are implemented on a novel servo-controlled swinging platform, essential for accurate real-time tracking in enclosed crowded areas. The outcomes of the research show that the proposed crowd-monitoring system can accurately localize an individual using LiDAR scan data in terms of his/her distance and angle from an image with a bounding box aiming to classify the detected object as a human being with high accuracy by using the proposed filtering and fusion techniques.
暂无评论