Search Results (5,120)

In producing English-Chinese bilingual maps, it is usually necessary to translate English place names into Chinese. However, pipeline-based methods for translating place names splits the place name translation task into multiple sub-tasks, carries the risk of error propagation, resulting in lower efficiency and poorer accuracy. Meanwhile, there is relatively little research on place name joint translation. In this regard, the study proposes an English-Chinese place name joint translation method based on prompt learning and knowledge graph enhancement. This method aims to improve the accuracy of English-Chinese place name translation. The proposed method is divided into two parts: The first part is the construction of prompt word template for place name translation. For the translation task of place names, the study first analyzes the characteristics of the transliteration of specific names and the semantic translation of generic names, constructing prompt word templates for the joint translation of ordinary place names. Then, based on the prompt words for ordinary place name translation, it takes into account the translation characteristics of the derived parts in derived place names, constructing a prompt word template for the joint translation of derived place names. Ultimately, leveraging the powerful contextual learning ability of LLM (Large Language Models), it achieves the joint translation of English and Chinese place names. The second part is the construction of the ontology of place name translation knowledge graph. To retrieve relevant knowledge about the input place names, the study designs an ontology for a knowledge graph of place names translation aimed at the English-Chinese place name translation task, combining the needs of English-Chinese place name translation and the semantic relationships between place names. This enhances the contextual information of the input place names and improves the performance of large language models in the English-Chinese place name translation task. Experiments have shown that compared to traditional pipeline-based place name translation methods, the place name translation method proposed in the study has improved performance by 21.26% in ordinary place name translation and an average of 27.70% in the field of derived place name translation. In bilingual map production, the method effectively improves the efficiency and accuracy of toponymic translation. Simultaneously providing reference for place name translation tasks in other languages. Full article

The mental health of university students has received much attention due to the various pressures of studies, life, and employment. Several studies have confirmed that campus public spaces contain multiple restorative potentials. Yet, the campus public space is still not ready to meet students’ new need for restorative percetions. Renewal practices for campus public spaces that integrate multi-issues are becoming more important, and further clarification of the measurement methods and optimization pathways is also needed. This study applied the semantic segmentation technique of the deep learning model to extract the feature indicators of outdoor public space based on street view image (SVI) data. The subjective evaluation of small-scale SVIs was obtained using the perceived restorative scale-11 (PRS-11) questionnaire. On this basis, restorative benefit evaluation models were established, including the explanatory and predictive models. The explanatory model used Pearson’s correlation and multiple linear regression analysis to identify the key indicators affecting restorative benefits, and the predictive model used the XGBoost 1.7.3 algorithm to predict the restorative benefit scores on the campus scale. The accessibility results from sDNA were then overlayed to form a comprehensive assessment matrix of restoration benefits and accessibility dimensions to identify further “areas with optimization potential”. In this way, three types of spatial dimensions (LRB-HA, HRB-LA, and LRB-LA) and sequential orders of temporal dimensions (short-term, medium-term, and long-term) were combined to propose optimization pathways for campus public space with the dual control of restorative benefits and accessibility. This study provides methodological guidelines and empirical data for campus regeneration and promotes outdoor public space efficiency. In addition, it can offer positive references for neighborhood-scale urban design and sustainable development. Full article

In real-time semantic segmentation for drone imagery, current lightweight algorithms suffer from the lack of integration of global and local information in the image, leading to missed detections and misclassifications in the classification categories. This paper proposes a method for the real-time semantic segmentation of drones that integrates multi-scale global context information. The principle utilizes a UNet structure, with the encoder employing a Resnet18 network to extract features. The decoder incorporates a global–local attention module, where the global branch compresses and extracts global information in both vertical and horizontal directions, and the local branch extracts local information through convolution, thereby enhancing the fusion of global and local information in the image. In the segmentation head, a shallow-feature fusion module is used to multi-scale integrate the various features extracted by the encoder, thereby strengthening the spatial information in the shallow features. The model was tested on the UAvid and UDD6 datasets, achieving accuracies of 68% mIoU (mean Intersection over Union) and 67% mIoU on the two datasets, respectively, 10% and 21.2% higher than the baseline model UNet. The real-time performance of the model reached 72.4 frames/s, which is 54.4 frames/s higher than the baseline model UNet. The experimental results demonstrate that the proposed model balances accuracy and real-time performance well. Full article

Object detection is essential for the perception systems of intelligent driving vehicles. RT-DETR has emerged as a prominent model. However, its direct application in intelligent driving vehicles still faces issues with the misdetection of occluded or small targets. To address these challenges, we propose a High-Precision Real-Time object detection algorithm (HPRT-DETR). We designed a Basic-iRMB-CGA (BIC) Block for a backbone network that efficiently extracts features and reduces the model’s parameters. We thus propose a Deformable Attention-based Intra-scale Feature Interaction (DAIFI) module by combining the Deformable Attention mechanism with the Intra-Scale Feature Interaction module. This enables the model to capture rich semantic features and enhance object detection accuracy in occlusion. The Local Feature Extraction Fusion (LFEF) block was created by integrating the local feature extraction module with the CNN-based Cross-scale Feature Fusion (CCFF) module. This integration expands the model’s receptive field and enhances feature extraction without adding learnable parameters or complex computations, effectively minimizing missed detections of small targets. Experiments on the KITTI dataset show that, compared to RT-DETR, HPRT-DETR improves mAP50 and FPS by 1.98% and 15.25%, respectively. Additionally, its generalization ability is assessed on the SODA 10M dataset, where HPRT-DETR outperforms RT-DETR in most evaluation metrics, confirming the model’s effectiveness. Full article

Empathetic and coherent responses are critical in automated chatbot-facilitated psychotherapy. This study addresses the challenge of enhancing the emotional and contextual understanding of large language models (LLMs) in psychiatric applications. We introduce Emotion-Aware Embedding Fusion, a novel framework integrating hierarchical fusion and attention mechanisms to prioritize semantic and emotional features in therapy transcripts. Our approach combines multiple emotion lexicons, including NRC Emotion Lexicon, VADER, WordNet, and SentiWordNet, with state-of-the-art LLMs such as Flan-T5, Llama 2, DeepSeek-R1, and ChatGPT 4. Therapy session transcripts, comprising over 2000 samples, are segmented into hierarchical levels (word, sentence, and session) using neural networks, while hierarchical fusion combines these features with pooling techniques to refine emotional representations. Attention mechanisms, including multi-head self-attention and cross-attention, further prioritize emotional and contextual features, enabling the temporal modeling of emotional shifts across sessions. The processed embeddings, computed using BERT, GPT-3, and RoBERTa, are stored in the Facebook AI similarity search vector database, which enables efficient similarity search and clustering across dense vector spaces. Upon user queries, relevant segments are retrieved and provided as context to LLMs, enhancing their ability to generate empathetic and contextually relevant responses. The proposed framework is evaluated across multiple practical use cases to demonstrate real-world applicability, including AI-driven therapy chatbots. The system can be integrated into existing mental health platforms to generate personalized responses based on retrieved therapy session data. The experimental results show that our framework enhances empathy, coherence, informativeness, and fluency, surpassing baseline models while improving LLMs’ emotional intelligence and contextual adaptability for psychotherapy. Full article

The threat posed by adversarial examples (AEs) to deep learning applications has garnered significant attention from the academic community. In response, various defense strategies have been proposed, including adversarial example detection. A range of detection algorithms has been developed to differentiate between benign samples and adversarial examples. However, the detection accuracy of these algorithms is significantly influenced by the characteristics of the adversarial attacks, such as attack type and intensity. Furthermore, the impact of image preprocessing on detection robustness—a common step before adversarial example generation—has been largely overlooked in prior research. To address these challenges, this paper introduces a novel adversarial example detection algorithm based on high-level feature differences (HFDs), which is specifically designed to improve robustness against both attacks and preprocessing operations. For each test image, a counterpart image with the same predicted label is randomly selected from the training dataset. The high-level features of both images are extracted using an encoder and compared through a similarity measurement model. If the feature similarity is low, the test image is classified as an adversarial example. The proposed method was evaluated for detection accuracy against four comparison methods, showing significant improvements over FS, DF, and MD, with a performance comparable to ESRM. Therefore, the subsequent robustness experiments focused exclusively on ESRM. Our results demonstrate that the proposed method exhibits superior robustness against preprocessing operations, such as downsampling and common corruptions, applied by attackers before generating adversarial examples. It is also applicable to various target models. By exploiting semantic conflicts in high-level features between clean and adversarial examples with the same predicted label, the method achieves high detection accuracy across diverse attack types while maintaining resilience to preprocessing, providing a valuable new perspective in the design of adversarial example detection algorithms. Full article

In this essay, we examine how the polyvalence of meaning in Kafka’s texts is engineered both semantically (on the narrative level) and syntactically (on the linguistic level), and we ask whether a computational approach can shed new light on the long-standing debate about the major characteristics of Kafka’s literary style. A mixed-method approach means that we seek out points of connection that interlink traditional humanist (i.e., interpretative) and computational (i.e., quantitative) methods of investigation. Following the introduction, the second section of our article provides a critical overview of the existing scholarship from both a humanist and a computational perspective. We argue that the main methodological difference between traditional humanist and AI-enhanced computational studies of Kafka’s literary style lies not in the use of statistics but in the new interpretative possibilities enabled by AI methods to explore stylistic features beyond the scope of human comprehension. In the third and fourth sections of our article, we will introduce our own stylometric approach to Kafka, detail our methods, and interpret our findings. Rather than focusing on training an AI model capable of accurately attributing authorship to Kafka, we examine whether AI could help us detect significant stylistic differences between the writing Kafka himself published during his lifetime (Kafka Core) and his posthumous writings edited and published by Max Brod. Full article

Effectively handling mixed noise types and varying intensities is crucial for accurate information extraction and analysis, particularly in resource-limited edge computing scenarios. Conventional image denoising approaches struggle with unseen noise distributions, limiting their effectiveness in real-world applications such as object detection, classification, and change detection. To address these challenges, we introduce a novel image denoising framework that integrates asymmetric learning with symmetric fusion. It leverages a pretrained model trained only on clean images to provide semantic priors, while a supervised module learns direct noise-to-clean mappings using paired noisy–clean data. The asymmetry in our approach stems from its dual training objectives: a pretrained encoder extracts semantic priors from noise-free data, while a supervised module learns noise-to-clean mappings. The symmetry is achieved through a structured fusion of pretrained priors and supervised features, enhancing generalization across diverse noise distributions, including those in edge computing environments. Extensive evaluations across multiple noise types and intensities, including real-world remote sensing data, demonstrate the superior robustness of our approach. Our method achieves state-of-the-art performance in both in-distribution and out-of-distribution noise scenarios, significantly enhancing image quality for downstream tasks such as environmental monitoring and disaster response. Future work may explore extending this framework to specialized applications like hyperspectral imaging and nighttime analysis while further refining the interplay between symmetry and asymmetry in deep-learning-based image restoration. Full article

Semantic segmentation is a crucial task in the field of computer vision, with important applications in areas such as autonomous driving, medical image analysis, and remote sensing image analysis. Dual-branch and multi-branch semantic segmentation networks that leverage deep learning technologies can enhance both segmentation accuracy and speed. These networks typically contain a semantic branch and a context branch. However, the feature maps in the detail branch are limited to a single type of receptive field, which limits models’ abilities to perceive objects at different scales. During the feature map fusion process, low-resolution feature maps from the semantic branch are upsampled with a large factor to match the feature maps in the detail branch. Unfortunately, these upsampling operations inevitably introduce noise. To address these issues, we propose several improvements to optimize the detail and semantic branches. We first design a receptive field-driven feature enhancement module to enrich the receptive fields of feature maps in the detail branch. Then, we propose a stepwise upsampling and fusion module to reduce the noise introduced during the upsampling process of feature fusion. Finally, we introduce a pyramid mixed pooling module (PMPM) to improve models’ abilities to perceive objects of different shapes. Considering the diversity of objects in terms of scale, shape, and category in urban street scene data, we carried out experiments on the Cityscapes and CamVid datasets. The experimental results on both datasets validate the effectiveness and efficiency of the proposed improvements. Full article

Named entity recognition in online medical consultation aims to address the challenge of identifying various types of medical entities within complex and unstructured social text in the context of online medical consultations. This can provide important data support for constructing more powerful online medical consultation knowledge graphs and improving virtual intelligent health assistants. A dataset of 26 medical entity types for named entity recognition for online medical consultations is first constructed. Then, a novel approach for deep named entity recognition in the medical field based on the fusion context mechanism is proposed. This approach captures enhanced local and global contextual semantic representations of online medical consultation text while simultaneously modeling high- and low-order feature interactions between local and global contexts, thereby effectively improving the sequence labeling performance. The experimental results show that the proposed approach can effectively identify 26 medical entity types with an average F1 score of 85.47%, outperforming the state-of-the-art (SOTA) method. The practical significance of this study lies in improving the efficiency and performance of domain-specific knowledge extraction in online medical consultation, supporting the development of virtual intelligent health assistants based on large language models and enabling real-time intelligent medical decision-making, thereby helping patients and their caregivers access common medical information more promptly. Full article

This study presents a practical methodology for integrating the multiscale spatial information of archaeological sites by combining Geographic Information Systems (GISs) with Historic Building Information Modelling (HBIM). The methodology categorises and integrates data based on its type and geometric scale, leveraging advanced 3D surveying techniques alongside semantic and parametric modelling tools. A multiscale system is proposed to manage heterogeneous geospatial data efficiently, enabling the development of enriched geometric models with detailed semantic and parametric attributes. The effectiveness of this approach is demonstrated through a case study of the Archaeological Area of Ancient “Abellinum”, showcasing seamless integration between HGISs and HBIM across multiple levels of detail. This work highlights the potential for enhanced management and the interpretation of archaeological heritage using innovative digital methodologies, highlighting the importance of representation in documenting historical transformations. Full article

Magnetic Resonance Angiography (MRA) is widely used for cerebrovascular assessment, with Time-of-Flight (TOF) MRA being a common non-contrast imaging technique. However, maximum intensity projection (MIP) images generated from TOF-MRA often include non-essential vascular structures such as external carotid branches, requiring manual editing for accurate visualization of intracranial arteries. This study proposes a deep learning-based semantic segmentation approach to automate the removal of these structures, enhancing MIP image clarity while reducing manual workload. Using DeepLab v3+, a convolutional neural network model optimized for segmentation accuracy, the method achieved an average Dice Similarity Coefficient (DSC) of 0.9615 and an Intersection over Union (IoU) of 0.9261 across five-fold cross-validation. The developed system processed MRA datasets at an average speed of 16.61 frames per second, demonstrating real-time feasibility. A dedicated software tool was implemented to apply the segmentation model directly to DICOM images, enabling fully automated MIP image generation. While the model effectively removed most external carotid structures, further refinement is needed to improve venous structure suppression. These results indicate that deep learning can provide an efficient and reliable approach for automated cerebrovascular image processing, with potential applications in clinical workflows and neurovascular disease diagnosis. Full article

To address the critical need for collision risk warning at unsignalized intersections, this study proposes an advanced predictive system combining YOLOv8 for object detection, Deep SORT for tracking, and Bi-LSTM networks for trajectory prediction. To adapt YOLOv8 for complex intersection scenarios, several architectural enhancements were incorporated. The RepLayer module replaced the original C2f module in the backbone, integrating large-kernel depthwise separable convolution to better capture contextual information in cluttered environments. The GIoU loss function was introduced to improve bounding box regression accuracy, mitigating the issues related to missed or incorrect detections due to occlusion and overlapping objects. Furthermore, a Global Attention Mechanism (GAM) was implemented in the neck network to better learn both location and semantic information, while the ReContext gradient composition feature pyramid replaced the traditional FPN, enabling more effective multi-scale object detection. Additionally, the CSPNet structure in the neck was substituted with Res-CSP, enhancing feature fusion flexibility and improving detection performance in complex traffic conditions. For tracking, the Deep SORT algorithm was optimized with enhanced appearance feature extraction, reducing the identity switches caused by occlusions and ensuring the stable tracking of vehicles, pedestrians, and non-motorized vehicles. The Bi-LSTM model was employed for trajectory prediction, capturing long-range dependencies to provide accurate forecasting of future positions. The collision risk was quantified using the predictive collision risk area (PCRA) method, categorizing risks into three levels (danger, warning, and caution) based on the predicted overlaps in trajectories. In the experimental setup, the dataset used for training the model consisted of 30,000 images annotated with bounding boxes around vehicles, pedestrians, and non-motorized vehicles. Data augmentation techniques such as Mosaic, Random_perspective, Mixup, HSV adjustments, Flipud, and Fliplr were applied to enrich the dataset and improve model robustness. In real-world testing, the system was deployed as part of the G310 highway safety project, where it achieved a mean Average Precision (mAP) of over 90% for object detection. Over a one-month period, 120 warning events involving vehicles, pedestrians, and non-motorized vehicles were recorded. Manual verification of the warnings indicated a prediction accuracy of 97%, demonstrating the system’s reliability in identifying potential collisions and issuing timely warnings. This approach represents a significant advancement in enhancing safety at unsignalized intersections in urban traffic environments. Full article

Hierarchical classification, which organizes items into structured categories and subcategories, has emerged as a powerful solution for handling large and complex datasets. However, traditional flat classification approaches often overlook the hierarchical dependencies between classes, leading to suboptimal predictions and limited interpretability. This paper addresses these challenges by proposing a novel integration of tree-based models with hierarchical-aware split criteria through adjusted entropy calculations. The proposed method calculates entropy at multiple hierarchical levels, ensuring that the model respects the taxonomic structure during training. This approach aligns statistical optimization with class semantic relationships, enabling more accurate and coherent predictions. Experiments conducted on real-world datasets structured according to the GS1 Global Product Classification (GPC) system demonstrate the effectiveness of our method. The proposed model was applied using tree-based ensemble methods combined with the newly developed hierarchy-aware metric Penalized Information Gain (PIG). PIG was implemented with level-wise entropy adjustments, assigning greater weight to higher hierarchical levels to maintain the taxonomic structure. The model was trained and evaluated on two real-world datasets based on the GS1 Global Product Classification (GPC) system. The final dataset included approximately 30,000 product descriptions spanning four hierarchical levels. An 80-20 train–test split was used, with model hyperparameters optimized through 5-fold cross-validation and Bayesian search. The experimental results showed a 12.7% improvement in classification accuracy at the lowest hierarchy level compared to traditional flat classification methods, with significant gains in datasets featuring highly imbalanced class distributions and deep hierarchies. The proposed approach also increased the F1 score by 12.6%. Despite these promising results, challenges remain in scaling the model for very large datasets and handling classes with limited training samples. Future research will focus on integrating neural networks with hierarchy-aware metrics, enhancing data augmentation to address class imbalance, and developing real-time classification systems for practical use in industries such as retail, logistics, and healthcare. Full article

In the early stages of architectural design, architects convert initial ideas into concrete design schemes, which heavily rely on their creativity and consume considerable time. Therefore, generative design methods based on artificial intelligence are promising for such tasks. However, effectively communicating design concepts to machines is challenging. To address this challenge, this paper proposes a novel cross-model approach for architectural design concepts using textual descriptions to assist architects, comprising a design concept extraction module and an architectural appearance generation module. The design concept extraction module adopts a contrastive learning framework to yield a text encoder with semantic extraction. Subsequently, the architectural appearance generation module proposes a novel deep sparse and text fusion generative adversarial network to convert the extracted design concept semantics into conceptual sketches, utilizing the unique sparsity of sketches. Additionally, it employs the pre-trained latent stable diffusion model to generate realistic and diverse high-rise building renderings, simulating the recreation processes of architects. The generated designs are evaluated qualitatively and quantitatively and further compared with existing real-life buildings to demonstrate the effectiveness of the proposed method. Furthermore, we demonstrate the feasibility of applying the proposed methodology in the early stages of architectural design by modeling a generated design. Full article