Scene recovery is a crucial imaging task that holds significant relevance in various practical domains, such as video surveillance and autonomous vehicles, among others. To improve the visual quality of sand-dust imag...
详细信息
Ensuring data security and privacy in Internet of Things (IoT) is increasingly critical due to the growing interconnectedness of devices and the sensitivity of the data they handle. This paper presents a novel approac...
详细信息
Person re-identification (Re-ID) aims to accurately match individuals across different camera views, a critical task for surveillance and security applications, often under varying conditions such as illumination, pos...
详细信息
Person re-identification (Re-ID) aims to accurately match individuals across different camera views, a critical task for surveillance and security applications, often under varying conditions such as illumination, pose, and background. Traditional Re-ID systems operate solely in the visible spectrum, which limits their effectiveness under varying lighting conditions and at night. To overcome these limitations, leveraging the visible-infrared (VIS-IR) domain becomes essential, as infrared imaging provides reliable information in low-light and night-time environments. However, the integration of VIS (visible) and IR (infrared) modalities introduces significant cross-modality discrepancies, posing a major challenge for feature alignment and fusion. To address this, we propose NiCTRAM: a Nyströmformer-based Cross-Modality Transformer designed for robust VIS-IR person re-identification. Our framework begins by extracting hierarchical features from both RGB and IR images through a shared convolutional neural network (CNN) backbone, ensuring the preservation of modality-specific characteristics. These features are then processed by parallel Nyströmformer encoders, which efficiently capture long-range dependencies in linear time using lightweight self-attention mechanisms. To bridge the modality gap, a cross-attention fusion block is introduced, where RGB and IR features interact and integrate second-order covariance statistics to model higher-order correlations. The fused features are subsequently refined through projection layers and optimized for re-identification using a classification head. Extensive experiments on benchmark VIS-IR person Re-ID datasets demonstrate that NiCTRAM outperforms existing methods, achieving state-of-the-art accuracy and robustness by effectively addressing the cross-modality challenges inherent in VIS-IR Re-ID. The proposed NiCTRAM model achieves significant improvements over the current SOTA in VIS-IR ReID. On the SYSU-MM01 dataset, it surpa
A traditional power grid integrates a Smart Grid (SG) technology to reduce greenhouse gases and increase the efficiency of energy transition. The Vehicle to Grid (V2G) is raised and combined with the SG network to ful...
详细信息
A traditional power grid integrates a Smart Grid (SG) technology to reduce greenhouse gases and increase the efficiency of energy transition. The Vehicle to Grid (V2G) is raised and combined with the SG network to fulfill these objectives of the SG technology. The two-way power flow in the V2G technology allows an Electrical Vehicle (EV) to charge its battery and discharge surplus energy back to the power grid through the Charging Stations (CSs). During the energy transfer, an EV shares identity, location, and charging preferences with the CS through an insecure channel. It raises significant security and privacy vulnerabilities for the V2G network. In addition, the EV and CS are situated in an exposed location that may increase the risk of physical attack. Hence, there is a need to preserve the security and privacy of the EV and CS in the V2G network. Moreover, a lightweight security solution is necessary for the resource constrained CS and EV in the V2G network. Several authentication and key agreement protocols were suggested in the literature to overcome the security challenges in the V2G network. However, the existing approaches fail to maintain the session key secrecy and preserve from the physical attack. Thus, we propose a Physically Unclonable Function (PUF) based Privacy Preserving Authentication and Key Agreement (P3AKA) framework for the V2G network using lightweight cryptographic operations. PUF protects the CS and EV from the physical attack and other lightweight cryptographic functions safeguard the network from other security attacks. Security analysis of the proposed P3AKA framework represents that it protects the V2G network from potential security threats such as impersonation, replay, Man-in-The-Middle (MiTM), physical, and machine learning attacks. Further, it ensures user anonymity and non-traceability of the EV user. The formal security verification uses the ROR model and Scyther tool to verify the proposed P3AKA framework. It illustrates that
Partially Observable Markov Decision Processes (POMDPs) provide a robust framework for decision-making under uncertainty in applications such as autonomous driving and robotic exploration. Their extension, ρPOMDPs, i...
详细信息
Analysis of medical images plays a vital role in the early detection and prognosis of diseases. Among various medical imaging, Magnetic Resonance Imaging(MRI) is valuable in providing insights into the brain artefact....
详细信息
Analysis of medical images plays a vital role in the early detection and prognosis of diseases. Among various medical imaging, Magnetic Resonance Imaging(MRI) is valuable in providing insights into the brain artefact. Early classification of brain tumors has become indispensable for doctors and medical persons. With the recent advancements in deep learning, Convolution Neural Networks have gained widespread in the domain of disease diagnosis and medical imaging, due to their notable efficacy and high performance. In this study, we have proposed the DiagPCNN model which incorporates a pre-trained model with CNN for an increment in optimal efficacy of the manifest model. We have also analysed the pros and cons of the DiagPCNN model over eighteen different architectures and three DiagPCNN models outperforms the rest of models with superior accuracy of 96.63, 97.28 and 97.79% respectively. Hence, simplifying their skilfulness in detecting brain tumors to induce the manifest model of a productive and virtuous approach for enhanced prognosis.
Recently, deep reinforcement learning (DRL) has emerged as a promising approach for robotic control. However, the deployment of DRL in real-world robots is hindered by its sensitivity to environmental perturbations. W...
详细信息
In real-world industrial scenarios, fault detection faces the widely recognized challenge of data imbalance, which not only refers to the scarcity of fault data but also includes the imbalance in healthy data. This ar...
详细信息
A convolutional neural network (CNN)-based face emotion recognition system that analyzes human emotions in real-time, including happy, sorrow, rage, and surprise, is described. Existing approaches often struggle with ...
详细信息
ISBN:
(数字)9798331527518
ISBN:
(纸本)9798331527525
A convolutional neural network (CNN)-based face emotion recognition system that analyzes human emotions in real-time, including happy, sorrow, rage, and surprise, is described. Existing approaches often struggle with accurately recognizing subtle or complex emotions due to limited feature extraction capabilities and the need for large, diverse training datasets. Additionally, they face challenges in adapting to real-world conditions, such as variations in lighting, angles, and occlusions. In order to improve the model's generalizability, the proposed research used a preprocessing pipeline that included gray scale conversion, scaling, normalization, and augmentation techniques including rotation and flipping on a labeled and balanced dataset of facial photos. Dropout and regularization techniques were used to reduce over fitting, and the CNN architecture was meticulously adjusted to maximize speed. High accuracy and resilience across emotion categories are demonstrated by preliminary results, indicating the model's potential for real-world uses in security surveillance, healthcare, and human-computer interaction.
This paper proposes a hybrid AI model for QoS estimation in 6G networks by mitigating the challenges caused due to α-η-μ fading and co-channel interference. These incorporate a variety of machine learning strategie...
详细信息
ISBN:
(数字)9798331543891
ISBN:
(纸本)9798331543907
This paper proposes a hybrid AI model for QoS estimation in 6G networks by mitigating the challenges caused due to α-η-μ fading and co-channel interference. These incorporate a variety of machine learning strategies that deliver improved accuracy with respect to forecasting and real-time updating capabilities. We evaluate it using an extensive set of simulations, including heatmaps and time series analysis; we show substantial reductions on system throughput, latency and utilization. This model makes the design highly dynamic to accommodate a wide spectrum of network conditions and thereby optimizes resources utilization in the overall performance. The comparison proves the high performance of hybrid AI model compared to conventional and state-of-the-art techniques, making it a promising resolution for QoS estimation in advanced network environments.
暂无评论