Search Results (22)

Federated learning (FL) has emerged as a powerful approach for privacy-preserving model training in autonomous vehicle networks, where real-world deployments rely on multiple roadside units (RSUs) serving heterogeneous clients with intermittent connectivity. While most research focuses on single-server or hierarchical cloud-based FL, multi-server FL can alleviate the communication bottlenecks of traditional setups. To this end, we propose an edge-based, multi-server FL (MS-FL) framework that combines performance-driven aggregation at each server—including statistical weighting of peer updates and outlier mitigation—with an application layer handover protocol that preserves model updates when vehicles move between RSU coverage areas. We evaluate MS-FL on both MNIST and GTSRB benchmarks under shard- and Dirichlet-based non-IID splits, comparing it against single-server FL and a two-layer edge-plus-cloud baseline. Over multiple communication rounds, MS-FL with the Statistical Performance-Aware Aggregation method and Dynamic Weighted Averaging Aggregation achieved up to a 20-percentage-point improvement in accuracy and consistent gains in precision, recall, and F1-score (95% confidence), while matching the low latency of edge-only schemes and avoiding the extra model transfer delays of cloud-based aggregation. These results demonstrate that coordinated cooperation among servers based on model quality and seamless handovers can accelerate convergence, mitigate data heterogeneity, and deliver robust, privacy-aware learning in connected vehicle environments. Full article

The increasing deployment of autonomous vehicles (AVs) has exposed critical vulnerabilities in traffic sign classification systems, particularly against adversarial attacks that can compromise safety. This study proposes a dual-purpose defense framework based on convolutional autoencoders to enhance robustness against two prominent white-box attacks: Fast Gradient Sign Method (FGSM) and Projected Gradient Descent (PGD). Experiments on the German Traffic Sign Recognition Benchmark (GTSRB) dataset show that, although these attacks can significantly degrade system performance, the proposed models are capable of partially recovering lost accuracy. Notably, the defense demonstrates strong capabilities in both detecting and reconstructing manipulated traffic signs, even under low-perturbation scenarios. Additionally, a feature-based autoencoder is introduced, which—despite a high false positive rate—achieves perfect detection in critical conditions, a tradeoff considered acceptable in safety-critical contexts. These results highlight the potential of autoencoder-based architectures as a foundation for resilient AV perception while underscoring the need for hybrid models integrating visual-language frameworks for real-time, fail-safe operation. Full article

This paper presents a traffic sign recognition (TSR) system based on the deep convolutional neural network (CNN) architecture, which proves to be extremely accurate in recognizing traffic signs under challenging conditions such as bad weather, low-resolution images, and various environmental-impact factors. The proposed CNN is compared with other architectures, including GoogLeNet, AlexNet, DarkNet-53, ResNet-34, VGG-16, and MicronNet-BF. Experimental results confirm that the proposed CNN significantly improves recognition accuracy compared to existing models. In order to make our model interpretable, we utilize explainable AI (XAI) approaches, specifically Gradient-weighted Class Activation Mapping (Grad-CAM), that can give insight into how the system comes to its decision. The evaluation of the Tsinghua-Tencent 100K (TT100K) traffic sign dataset showed that the proposed method significantly outperformed existing state-of-the-art methods. Additionally, we evaluated our model on the German Traffic Sign Recognition Benchmark (GTSRB) dataset to ensure generalization, demonstrating its ability to perform well in diverse traffic sign conditions. Design issues such as noise, contrast, blurring, and zoom effects were added to enhance performance in real applications. These verified results indicate both the strength and reliability of the CNN architecture proposed for TSR tasks and that it is a good option for integration into intelligent transportation systems (ITSs). Full article

Traffic safety remains a pressing global concern, with traffic signs playing a vital role in regulating and guiding drivers. However, environmental factors like lighting and weather often compromise their visibility, impacting human drivers and autonomous vehicle (AV) systems. This study addresses critical traffic sign detection (TSD) and classification (TSC) gaps by leveraging the YOLOv8 algorithm to evaluate the detection accuracy and sign quality under diverse lighting conditions. The model achieved robust performance metrics across day and night scenarios using the novel ZND dataset, comprising 16,500 labeled images sourced from the GTSRB, GitHub repositories, and real-world own photographs. Complementary retroreflectivity assessments using handheld retroreflectometers revealed correlations between the material properties of the signs and their detection performance, emphasizing the importance of the retroreflective quality, especially under night-time conditions. Additionally, video analysis highlighted the influence of sharpness, brightness, and contrast on detection rates. Human evaluations further provided insights into subjective perceptions of visibility and their relationship with algorithmic detection, underscoring areas for potential improvement. The findings emphasize the need for using various assessment methods, advanced algorithms, enhanced sign materials, and regular maintenance to improve detection reliability and road safety. This research bridges the theoretical and practical aspects of TSD, offering recommendations that could advance AV systems and inform future traffic sign design and evaluation standards. Full article

Kolmogorov–Arnold Networks (KANs) are a novel class of neural network architectures based on the Kolmogorov–Arnold representation theorem, which has demonstrated potential advantages in accuracy and interpretability over Multilayer Perceptron (MLP) models. This paper comprehensively evaluates the robustness of various KAN architectures—including KAN, KAN-Mixer, KANConv_KAN, and KANConv_MLP—against adversarial attacks, which constitute a critical aspect that has been underexplored in current research. We compare these models with MLP-based architectures such as MLP, MLP-Mixer, and ConvNet_MLP across three traffic sign classification datasets: GTSRB, BTSD, and CTSD. The models were subjected to various adversarial attacks (FGSM, PGD, CW, and BIM) with varying perturbation levels and were trained under different strategies, including standard training, adversarial training, and Randomized Smoothing. Our experimental results demonstrate that KAN-based models, particularly the KAN-Mixer, exhibit superior robustness to adversarial attacks compared to their MLP counterparts. Specifically, the KAN-Mixer consistently achieved lower Success Attack Rates (SARs) and Degrees of Change (DoCs) across most attack types and datasets while maintaining high accuracy on clean data. For instance, under FGSM attacks with $ϵ=0.01$, the KAN-Mixer outperformed the MLP-Mixer by maintaining higher accuracy and lower SARs. Adversarial training and Randomized Smoothing further enhanced the robustness of KAN-based models, with t-SNE visualizations revealing more stable latent space representations under adversarial perturbations. These findings underscore the potential of KAN architectures to improve neural network security and reliability in adversarial settings. Full article

Accurately classifying road signs is crucial for autonomous driving due to the high stakes involved in ensuring safety and compliance. As Convolutional Neural Networks (CNNs) have largely replaced traditional Machine Learning models in this domain, the demand for substantial training data has increased. This study aims to compare the performance of classical Machine Learning (ML) models and Deep Learning (DL) models under varying amounts of training data, particularly focusing on altered signs to mimic real-world conditions. We evaluated three classical models: Support Vector Machine (SVM), Random Forest, and Linear Discriminant Analysis (LDA), and one Deep Learning model: Convolutional Neural Network (CNN). Using the German Traffic Sign Recognition Benchmark (GTSRB) dataset, which includes approximately 40,000 German traffic signs, we introduced digital alterations to simulate conditions such as environmental wear or vandalism. Additionally, the Histogram of Oriented Gradients (HOG) descriptor was used to assist classical models. Bayesian optimization and k-fold cross-validation were employed for model fine-tuning and performance assessment. Our findings reveal a threshold in training data beyond which accuracy plateaus. Classical models showed a linear performance decrease under increasing alteration, while CNNs, despite being more robust to alterations, did not significantly outperform classical models in overall accuracy. Ultimately, classical Machine Learning models demonstrated performance comparable to CNNs under certain conditions, suggesting that effective road sign classification can be achieved with less computationally intensive approaches. Full article

Efficient traffic sign recognition is crucial to facilitating the intelligent driving of new energy vehicles. However, current approaches like the Vision Transformer (ViT) model often impose high storage and computational demands, escalating hardware costs. This paper presents a similarity filter pruning method based on locally linear embedding. Using the alternating direction multiplier method and the loss of the locally linear embedding method for the model training function, the proposed pruning method prunes the operation model mainly by evaluating the similarity of each layer in the network layer filters. According to the pre-set pruning threshold value, similar filters to be pruned are obtained, and the filter with a large cross-entropy value is retained. The results from the Belgium Traffic Sign (BelgiumTS) and German Traffic Sign Recognition Benchmark (GTSRB) datasets indicate that the proposed similarity filter pruning based on local linear embedding (SJ-LLE) pruning algorithm can reduce the number of parameters of the multi-head self-attention module and Multi-layer Perceptron (MLP) module of the ViT model by more than 60%, and the loss of model accuracy is acceptable. The scale of the ViT model is greatly reduced, which is conducive to applying this model in embedded traffic sign recognition equipment. Also, this paper proves the hypothesis through experiments that “using the LLE algorithm as the loss function for model training before pruning plays a positive role in reducing the loss of model performance in the pruning process”. Full article

Traffic Sign Recognition (TSR) plays a vital role in intelligent transportation systems (ITS) to improve road safety and optimize traffic management. While existing TSR models perform well in challenging scenarios, their lack of transparency and interpretability hinders reliability, trustworthiness, validation, and bias identification. To address this issue, we propose a Convolutional Neural Network (CNN)-based model for TSR and evaluate its performance on three benchmark datasets: German Traffic Sign Recognition Benchmark (GTSRB), Indian Traffic Sign Dataset (ITSD), and Belgian Traffic Sign Dataset (BTSD). The proposed model achieves an accuracy of 98.85% on GTSRB, 94.73% on ITSD, and 92.69% on BTSD, outperforming several state-of-the-art frameworks, such as VGG19, VGG16, ResNet50V2, MobileNetV2, DenseNet121, DenseNet201, NASNetMobile, and EfficientNet, while also providing faster training and response times. We further enhance our model by incorporating explainable AI (XAI) techniques, specifically, Local Interpretable Model-Agnostic Explanations (LIME) and Gradient-weighted Class Activation Mapping (Grad-CAM), providing clear insights of the proposed model decision-making process. This integration allows the extension of our TSR model to various engineering domains, including autonomous vehicles, advanced driver assistance systems (ADAS), and smart traffic control systems. The practical implementation of our model ensures real-time, accurate recognition of traffic signs, thus optimizing traffic flow and minimizing accident risks. Full article

Artificial Intelligence (AI) in the automotive industry allows car manufacturers to produce intelligent and autonomous vehicles through the integration of AI-powered Advanced Driver Assistance Systems (ADAS) and/or Automated Driving Systems (ADS) such as the Traffic Sign Recognition (TSR) system. Existing TSR solutions focus on some categories of signs they recognise. For this reason, a TSR approach encompassing more road sign categories like Warning, Regulatory, Obligatory, and Priority signs is proposed to build an intelligent and real-time system able to analyse, detect, and classify traffic signs into their correct categories. The proposed approach is based on an overview of different Traffic Sign Detection (TSD) and Traffic Sign Classification (TSC) methods, aiming to choose the best ones in terms of accuracy and processing time. Hence, the proposed methodology combines the Haar cascade technique with a deep CNN model classifier. The developed TSC model is trained on the GTSRB dataset and then tested on various categories of road signs. The achieved testing accuracy rate reaches 98.56%. In order to improve the classification performance, we propose a new attention-based deep convolutional neural network. The achieved results are better than those existing in other traffic sign classification studies since the obtained testing accuracy and F1-measure rates achieve, respectively, 99.91% and 99%. The developed TSR system is evaluated and validated on a Raspberry Pi 4 board. Experimental results confirm the reliable performance of the suggested approach. Full article

Recognizing and classifying traffic signs is a challenging task that can significantly improve road safety. Deep neural networks have achieved impressive results in various applications, including object identification and automatic recognition of traffic signs. These deep neural network-based traffic sign recognition systems may have limitations in practical applications due to their computational requirements and resource consumption. To address this issue, this paper presents a lightweight neural network for traffic sign recognition that achieves high accuracy and precision with fewer trainable parameters. The proposed model is trained on the German Traffic Sign Recognition Benchmark (GTSRB) and Belgium Traffic Sign (BelgiumTS) datasets. Experimental results demonstrate that the proposed model has achieved 98.41% and 92.06% accuracy on GTSRB and BelgiumTS datasets, respectively, outperforming several state-of-the-art models such as GoogleNet, AlexNet, VGG16, VGG19, MobileNetv2, and ResNetv2. Furthermore, the proposed model outperformed these methods by margins ranging from 0.1 to 4.20 percentage point on the GTSRB dataset and by margins ranging from 9.33 to 33.18 percentage point on the BelgiumTS dataset. Full article

With the growing trend in autonomous vehicles, accurate recognition of traffic signs has become crucial. This research focuses on the use of convolutional neural networks for traffic sign classification, specifically utilizing pre-trained models of ResNet50, DenseNet121, and VGG16. To enhance the accuracy and robustness of the model, the authors implement an ensemble learning technique with majority voting, to combine the predictions of multiple CNNs. The proposed approach was evaluated on three different traffic sign datasets: the German Traffic Sign Recognition Benchmark (GTSRB), the Belgium Traffic Sign Dataset (BTSD), and the Chinese Traffic Sign Database (TSRD). The results demonstrate the efficacy of the ensemble approach, with recognition rates of 98.84% on the GTSRB dataset, 98.33% on the BTSD dataset, and 94.55% on the TSRD dataset. Full article

Quantum machine learning (QML) has attracted significant research attention over the last decade. Multiple models have been developed to demonstrate the practical applications of the quantum properties. In this study, we first demonstrate that the previously proposed quanvolutional neural network (QuanvNN) using a randomly generated quantum circuit improves the image classification accuracy of a fully connected neural network against the Modified National Institute of Standards and Technology (MNIST) dataset and the Canadian Institute for Advanced Research 10 class (CIFAR-10) dataset from 92.0% to 93.0% and from 30.5% to 34.9%, respectively. We then propose a new model referred to as a Neural Network with Quantum Entanglement (NNQE) using a strongly entangled quantum circuit combined with Hadamard gates. The new model further improves the image classification accuracy of MNIST and CIFAR-10 to 93.8% and 36.0%, respectively. Unlike other QML methods, the proposed method does not require optimization of the parameters inside the quantum circuits; hence, it requires only limited use of the quantum circuit. Given the small number of qubits and relatively shallow depth of the proposed quantum circuit, the proposed method is well suited for implementation in noisy intermediate-scale quantum computers. While promising results were obtained by the proposed method when applied to the MNIST and CIFAR-10 datasets, a test against a more complicated German Traffic Sign Recognition Benchmark (GTSRB) dataset degraded the image classification accuracy from 82.2% to 73.4%. The exact causes of the performance improvement and degradation are currently an open question, prompting further research on the understanding and design of suitable quantum circuits for image classification neural networks for colored and complex data. Full article

Deep neural networks (DNNs) have been used successfully for many image classification problems. One of the most important factors that determines the final efficiency of a DNN is the correct construction of the training set. Erroneously labeled training images can degrade the final accuracy and additionally lead to unpredictable model behavior, reducing reliability. In this paper, we propose MultiNET, a novel method for the automatic detection of noisy labels within image datasets. MultiNET is an adaptation of the current state-of-the-art confident learning method. In contrast to the original, our method aggregates the outputs of multiple DNNs and allows for the adjustment of detection sensitivity. We conduct an exhaustive evaluation, incorporating four widely used datasets (CIFAR10, CIFAR100, MNIST, and GTSRB), eight state-of-the-art DNN architectures, and a variety of noise scenarios. Our results demonstrate that MultiNET significantly outperforms the confident learning method. Full article

Capsule networks are a type of neural network that use the spatial relationship between features to classify images. By capturing the poses and relative positions between features, this network is better able to recognize affine transformation and surpass traditional convolutional neural networks (CNNs) when handling translation, rotation, and scaling. The stacked capsule autoencoder (SCAE) is a state-of-the-art capsule network that encodes an image in capsules which each contain poses of features and their correlations. The encoded contents are then input into the downstream classifier to predict the image categories. Existing research has mainly focused on the security of capsule networks with dynamic routing or expectation maximization (EM) routing, while little attention has been given to the security and robustness of SCAEs. In this paper, we propose an evasion attack against SCAEs. After a perturbation is generated based on the output of the object capsules in the model, it is added to an image to reduce the contribution of the object capsules related to the original category of the image so that the perturbed image will be misclassified. We evaluate the attack using an image classification experiment on the Mixed National Institute of Standards and Technology Database (MNIST), Fashion-MNIST, and German Traffic Sign Recognition Benchmark (GTSRB) datasets, and the average attack success rate can reach 98.6%. The experimental results indicate that the attack can achieve high success rates and stealthiness. This finding confirms that the SCAE has a security vulnerability that allows for the generation of adversarial samples. Our work seeks to highlight the threat of this attack and focus attention on SCAE’s security. Full article

Embedding Artificial Intelligence onto low-power devices is a challenging task that has been partly overcome with recent advances in machine learning and hardware design. Presently, deep neural networks can be deployed on embedded targets to perform different tasks such as speech recognition, object detection or Human Activity Recognition. However, there is still room for optimization of deep neural networks onto embedded devices. These optimizations mainly address power consumption, memory and real-time constraints, but also an easier deployment at the edge. Moreover, there is still a need for a better understanding of what can be achieved for different use cases. This work focuses on quantization and deployment of deep neural networks onto low-power 32-bit microcontrollers. The quantization methods, relevant in the context of an embedded execution onto a microcontroller, are first outlined. Then, a new framework for end-to-end deep neural networks training, quantization and deployment is presented. This framework, called MicroAI, is designed as an alternative to existing inference engines (TensorFlow Lite for Microcontrollers and STM32Cube.AI). Our framework can indeed be easily adjusted and/or extended for specific use cases. Execution using single precision 32-bit floating-point as well as fixed-point on 8- and 16 bits integers are supported. The proposed quantization method is evaluated with three different datasets (UCI-HAR, Spoken MNIST and GTSRB). Finally, a comparison study between MicroAI and both existing embedded inference engines is provided in terms of memory and power efficiency. On-device evaluation is done using ARM Cortex-M4F-based microcontrollers (Ambiq Apollo3 and STM32L452RE). Full article