Search Results (20,545)

Accurate weld defect detection is critical for ensuring structural safety and evaluating welding quality in industrial applications. Manual inspection methods have inherent limitations, including inefficiency and inadequate sensitivity to subtle defects. Existing detection models, primarily designed for natural images, struggle to adapt to the characteristic challenges of weld X-ray images, such as high noise, low contrast, and inter-defect similarity, particularly leading to missed detections and false positives for small defects. To address these challenges, a multi-dimensional feature fusion model (MADet), which is a multi-branch deep fusion network for weld defect detection, was proposed. The framework incorporates two key innovations: (1) A multi-scale feature fusion network integrated with lightweight attention residual modules to enhance the perception of fine-grained defect features by leveraging low-level texture information. (2) An anchor-based feature-selective detection head was used to improve the discrimination and localization accuracy for five typical defect categories. Extensive experiments on both public and proprietary weld defect datasets demonstrated that MADet achieved significant improvements over the state-of-the-art YOLO variants. Specifically, it surpassed the suboptimal model by 7.41% in mAP@0.5, indicating strong industrial applicability. Full article

Existing studies on the abrasive cutting process have primarily focused on the influence of cutting conditions on key parameters such as temperature, cut-off wheel wear, and machined surface quality. However, the choice of working conditions is often made based on the experience of qualified personnel or using data from reference sources. The literature also provides optimal values for the cutting mode elements, but these are only valid for specific methods and cutting conditions. This article proposes a new multi-objective optimization approach for determining the conditions for the implementation of the abrasive cutting process that leads to Pareto-optimal solutions for improving economic efficiency, evaluated by production rate and manufacturing net cost parameters. To demonstrate this approach, the elastic abrasive cutting process of structural steels C45 and 42Cr4 has been selected. Theoretical–experimental models for production rate and manufacturing net cost have been developed, reflecting the complex influence of the conditions of the elastic abrasive cutting process (compression force of the cut-off wheel on the workpiece and rotational frequency of the workpiece). Multi-objective compromise optimization based on a genetic algorithm has been conducted by applying two methods—the determination of a compromise optimal area for the conditions of the elastic abrasive cutting process and the generalized utility function method. Optimal conditions for the implementation of the elastic abrasive cutting process have been determined, ensuring the best combination of high production rate and low manufacturing net cost. Full article

With the growing demand for ecological monitoring and geological exploration, high‑quality satellite remote‑sensing imagery has become indispensable for accurate information extraction and automated analysis. However, haze reduces image contrast and sharpness, significantly impairing quality. Existing dehazing methods, primarily designed for natural images, struggle with remote-sensing images due to their complex imaging conditions and scale diversity. Given this, we propose a novel Multi-Scale Contextual Attention Generative Adversarial Network (MCA-GAN), specifically designed for satellite image dehazing. Our method integrates multi‑scale feature extraction with global contextual guidance to enhance the network’s comprehension of complex remote‑sensing scenes and its sensitivity to fine details. MCA‑GAN incorporates two self-designed key modules: (1) a Multi‑Scale Feature Aggregation Block, which employs multi‑directional global pooling and multi‑scale convolutional branches to bolster the model’s ability to capture land‑cover details across varying spatial scales; (2) a Dynamic Contextual Attention Block, which uses a gated mechanism to fuse three‑dimensional attention weights with contextual cues, thereby preserving global structural and chromatic consistency while retaining intricate local textures. Extensive qualitative and quantitative experiments on public benchmarks demonstrate that MCA‑GAN outperforms other existing methods in both visual fidelity and objective metrics, offering a robust and practical solution for remote‑sensing image dehazing. Full article

Urban forests, as a form of green infrastructure, play a vital role in enhancing urban resilience, environmental health, and quality of life. However, users perceive and utilize these spaces in diverse ways. This study aims to identify latent perception types among urban forest visitors and analyze their behavioral, demographic, and policy-related characteristics in Incheon Metropolitan City (Republic of Korea). Using latent class analysis, four distinct visitor types were identified: multipurpose recreationists, balanced relaxation seekers, casual forest users, and passive forest visitors. Multipurpose recreationists preferred active physical use and sports facilities, while balanced relaxation seekers emphasized emotional well-being and cultural experiences. Casual users engaged lightly with forest settings, and passive forest visitors exhibited minimal recreational interest. Satisfaction with forest elements such as vegetation, facilities, and management conditions varied across visitor types and age groups, especially among older adults. These findings highlight the need for perception-based green infrastructure planning. Policy recommendations include expanding accessible neighborhood green spaces for aging populations, promoting community-oriented events, and offering participatory forest programs for youth engagement. By integrating user segmentation into urban forest planning and governance, this study contributes to more inclusive, adaptive, and sustainable management of urban green infrastructure. Full article

Accurate prediction of key ocean elements (e.g., chlorophyll-a concentration, sea surface temperature, etc.) is imperative for maintaining marine ecological balance, responding to marine disaster pollution, and promoting the sustainable use of marine resources. Existing spatio-temporal prediction models primarily rely on either physical or data-driven approaches. Physical models are constrained by modeling complexity and parameterization errors, while data-driven models lack interpretability and depend on high-quality data. To address these challenges, this study proposes OKG-ConvGRU, a domain knowledge-guided remote sensing prediction framework for ocean elements. This framework integrates knowledge graphs with the ConvGRU network, leveraging prior knowledge from marine science to enhance the prediction performance of ocean elements in remotely sensed images. Firstly, we construct a spatio-temporal knowledge graph for ocean elements (OKG), followed by semantic embedding representation for its spatial and temporal dimensions. Subsequently, a cross-attention-based feature fusion module (CAFM) is designed to efficiently integrate spatio-temporal multimodal features. Finally, these fused features are incorporated into an enhanced ConvGRU network. For multi-step prediction, we adopt a Seq2Seq architecture combined with a multi-step rolling strategy. Prediction experiments for chlorophyll-a concentration in the eastern seas of China validate the effectiveness of the proposed framework. The results show that, compared to baseline models, OKG-ConvGRU exhibits significant advantages in prediction accuracy, long-term stability, data utilization efficiency, and robustness. This study provides a scientific foundation and technical support for the precise monitoring and sustainable development of marine ecological environments. Full article

Background/Objectives: This study investigated the association between cochlear implant (CI) users’ assessed perception of musical sound quality and their subjective music perception and music-related quality of life (QoL). The aim was to provide a comprehensive evaluation by integrating a relatively objective Turkish Multiple Stimulus with Hidden Reference and Anchor (TR-MUSHRA) test and a subjective music questionnaire. Methods: Thirty CI users and thirty normal-hearing (NH) adults were assessed. Perception of sound quality was measured using the TR-MUSHRA test. Subjective assessments were conducted with the Music-Related Quality of Life Questionnaire (MuRQoL). Results: TR-MUSHRA results showed that while NH participants rated all filtered stimuli as perceptually different from the original, CI users provided similar ratings for stimuli with adjacent high-pass filter settings, indicating less differentiation in perceived sound quality. On the MuRQoL, groups differed on the Frequency subscale but not the Importance subscale. Critically, no significant correlation was found between the TR-MUSHRA scores and the MuRQoL subscale scores in either group. Conclusions: The findings demonstrate that TR-MUSHRA is an effective tool for assessing perceived sound quality relatively objectively, but there is no relationship between perceiving sound quality differences and measures of self-reported musical engagement and its importance. Subjective music experience may represent different domains beyond the perception of sound quality. Therefore, successful auditory rehabilitation requires personalized strategies that consider the multifaceted nature of music perception beyond simple perceptual judgments. Full article

Thermal analysis (TA) has been a valuable tool for controlling the carbon equivalent (CE) of cast irons. Additionally, this technique can provide enhanced control over melt quality, allowing for the avoidance of defects such as undesirable graphite morphology and the formation of carbides. To obtain the most valuable information from the TA, it is necessary to minimize the variations in the filling operation of the TA cups. However, the mass and pouring temperature of TA cups can vary in TA’s typical foundry operations. A design of experiments was performed to determine whether specific parameters of cooling curves used for quality control can distinguish the inoculation effect in the melt when the mass and the pouring temperature of TA cups are varied. The minimum temperature of the eutectic arrest proved to be a robust inoculation potential control parameter when variations in the cup’s mass were within a range of 268–390 g and were filled at any pouring temperature between 1235 and 1369 °C. Lighter cups under 268 g and poured at a low temperature are not suitable for controlling inoculation potential by TA; however, they remain helpful in controlling CE. These later cups are related to cooling times of less than 180 s, which can serve as a criterion for discarding unsuitable samples. A bimodal population of cell surfaces was revealed in the samples, with the population of small cells being proportionally more numerous in samples with lower TE_min values. Full article

Ensuring software quality during development requires effective regression testing. However, test suites in open-source libraries often grow large, redundant, and difficult to maintain. Most traditional test suite optimization methods treat test cases as atomic units, without analyzing the utility of individual instructions. This paper presents an intelligent method for test case synthesis using a Q-learning agent. The agent learns to construct compact test cases by interacting with an execution environment and receives rewards based on branch coverage improvements and simultaneous reductions in test case length. The training process includes a pretraining phase that transfers knowledge from the original test suite, followed by adaptive learning episodes on individual test cases. As a result, the method requires no formal documentation or API specifications and uses only execution traces of the original test cases. An explicit synthesis algorithm constructs new test cases by selecting API calls from a learned policy encoded in a Q-table. Experiments were conducted on two open-source C++ libraries of differing API complexity and original test suite size. The results show that the proposed method can reach up to 67% test suite reduction while preserving branch coverage, confirming its effectiveness for regression test suite minimization in resource-constrained or specification-limited environments. Full article

In the rock–soil–biology–water ecosystem, rock weathering provides essential plant nutrients. However, its supply is insufficient for rising crop demands under population growth and climate change, while excessive fertilizer causes soil degradation and pollution. This study innovatively irrigated with carbonate rock leachates to enhance soil nutrient availability. A pot experiment with lettuce showed that irrigation significantly increased soil NO₃⁻-N (+102.20%), available K (+16.45%), available P (+17.95%), Ca (+6.04%), Mg (+11.65%), and Fe (+11.60%), and elevated the relative abundance of Firmicutes. Lettuce biomass per plant rose by 23.78%, with higher leaf minerals (P, K, Ca, and Mg) and antioxidants (carotenoids and ascorbic acid). A field experiment further confirmed improvement of soil nutrient availability and peanut yield. This carbonate rock leachate irrigation technique effectively enhances soil quality and crop productivity/quality, offering a sustainable approach for green agriculture. Full article

The objective of this study was to evaluate the physical-hydric properties of a Planosol under an Integrated Crop–Livestock–Forest (ICLF) system in the Agreste region of Paraíba, Brazil, after eight years of implementation, and to compare them with areas under a conventional cropping system and secondary native vegetation. The experiment was conducted at the experimental station located in Alagoinha, in the Agreste mesoregion of the State of Paraíba, Brazil. The experimental design adopted was a randomized block design (RBD) with five treatments and four replications (5 × 4 + 2). The treatments consisted of: (1) Gliricidia (Gliricidia sepium (Jacq.) Steud) + Signal grass (Urochloa decumbens) (GL+SG); (2) Sabiá (Mimosa caesalpiniaefolia Benth) + Signal grass (SB+SG); (3) Purple Ipê (Handroanthus avellanedae (Lorentz ex Griseb.) Mattos) + SG (I+SG); (4) annual crop + SG (C+SG); and (5) Signal grass (SG). Two additional treatments were included for statistical comparison: a conventional cropping system (CC) and a secondary native vegetation area (NV), both located near the experimental site. The CC treatment showed the lowest bulk density (1.23 g cm⁻³) and the lowest degree of compaction (66.3%) among the evaluated treatments, as well as a total porosity (TP) higher than 75% (0.75 m³ m⁻³). In the soil under the integration system, the lowest bulk density (1.38 g cm⁻³) and the highest total porosity (0.48 m³ m⁻³) were observed in the SG treatment at the 0.0–0.10 m depth. High S-index values (>0.035) and a low relative field capacity (RFc < 0.50) and K_θ indicate high structural quality and low soil water storage capacity. It was concluded that the SG, I+SG, SB+SG, and CC treatments presented the highest values of soil bulk and degree of compaction in the layers below 0.10 m. The I+SG and C+SG treatments showed the lowest hydraulic conductivities and macroaggregation. The SG and C+SG treatments had the lowest available water content and available water capacity across the three analyzed soil layers. Full article

This work aims to solve the problem of product quality fluctuations caused by batch-to-batch variations in the adsorption capacity of activated carbon during the production of traditional Chinese medicine (TCM) injections. In this work, Shenqi Fuzheng injection was selected as an example. Diluted Shenqi Extract (DSE), an intermediate in the production process of Shenqi Fuzheng injection, was adsorbed with different batches of activated carbon. The adsorption capacities of adenine, adenosine, calycosin-7-glucoside, and astragaloside IV in DSE were selected as evaluation indices for activated carbon absorption. Characterization methods such as nitrogen adsorption, X-ray photoelectron spectrum (XPS), and Fourier transform infrared (FTIR) were chosen to explore the quantitative relationships between the properties of activated carbon (i.e., specific surface area, pore volume, surface elements, and spectrum) and the adsorption capacities of these four components. It was found that the characteristic wavelengths from FTIR characterization, i.e., 1560 cm⁻¹, 2325 cm⁻¹, 3050 cm⁻¹, and 3442 cm⁻¹, etc., showed the strongest correlation with the adsorption capacities of these four components. Prediction models based on the transmittance at characteristic wavelengths were successfully established via multiple linear regression. In validation experiments of models, the relative errors of predicted adsorption capacities of activated carbon were mostly within 5%, indicating good predictive ability of the models. The results of this work suggest that the prediction method of adsorption capacity based on the mid-infrared spectrum can provide a new way for the quality control of activated carbon. Full article

Interpreters who translate for refugees are regularly confronted with traumatic content. Listening regularly to and translating potentially traumatizing stories make interpreters vulnerable to secondary traumatization. The current study aimed to investigate secondary traumatic stress (STS) in interpreters working with refugees and to identify potential risk and protective factors. In this cross-sectional study, 64 interpreters from Austria, Germany, and Switzerland participated. STS, compassion satisfaction, burnout, and resilience were assessed using the Secondary Traumatic Stress Scale, Professional Quality of Life Scale, and Connor–Davidson Resilience Scale. Sociodemographic data were collected (e.g., experiences with translation in psychotherapy, personal backgrounds of forced displacement, and personal experiences with psychotherapy as a client). Subgroup comparisons were conducted to identify risk factors for STS. A total of 43 participants (67%) showed at least mild STS. STS was significantly associated with burnout. Personal experiences of forced displacement, gender, and working context had no impact on STS. Interpreters with personal experiences of psychotherapy showed a higher level of STS as well as higher resilience than those without personal experiences of psychotherapy. Independent of personal experiences of forced displacement, gender, and working context, interpreters who work with refugees are at high risk of STS. Regular training and supervision for interpreters who work with refugees should thus be offered as standard practice. Full article

In order to increase the effectiveness of model training, data reduction is essential to data-centric Artificial Intelligence (AI). It achieves this by locating the most instructive examples in massive datasets. To increase data quality and training efficiency, the main difficulty is choosing the best examples rather than the complete datasets. In this paper, we propose an effective data reduction strategy based on Pointwise 𝒱-Information (PVI). To enable a static method, we first use PVI to quantify instance difficulty and remove instances with low difficulty. Experiments show that classifier performance is maintained with only a 0.0001% to 0.76% decline in accuracy when 10–30% of the data is removed. Second, we train the classifiers using a progressive learning strategy on examples sorted by increasing PVI, accelerating convergence and achieving a 0.8% accuracy gain over conventional training. Our findings imply that training a classifier on the chosen optimal subset may improve model performance and increase training efficiency when combined with an efficient data reduction strategy. Furthermore, we have adapted the PVI framework, which was previously limited to English datasets, to a variety of Chinese Natural Language Processing (NLP) tasks and base models, yielding insightful results for faster training and cross-lingual data reduction. Full article

The accuracy of classification and localization of plants on images obtained from the board of an unmanned aerial vehicle (UAV) is of great importance when implementing precision farming technologies. It allows for the effective application of variable rate technologies, which not only saves chemicals but also reduces the environmental load on cultivated fields. Machine learning algorithms are widely used for plant classification. Research on the application of the YOLO algorithm is conducted for simultaneous identification, localization, and classification of plants. However, the quality of the algorithm significantly depends on the training set. The aim of this study is not only the detection of a cultivated plant (soybean) but also weeds growing in the field. The dataset developed in the course of the research allows for solving this issue by detecting not only soybean but also seven weed species common in the fields of Kazakhstan. The article describes an approach to the preparation of a training set of images for soybean fields using preliminary thresholding and bound box (Bbox) segmentation of marked images, which allows for improving the quality of plant classification and localization. The conducted research and computational experiments determined that Bbox segmentation shows the best results. The quality of classification and localization with the application of Bbox segmentation significantly increased (f1 score increased from 0.64 to 0.959, mAP50 from 0.72 to 0.979); for a cultivated plant (soybean), the best classification results known to date were achieved with the application of YOLOv8x on images obtained from the UAV, with an f1 score = 0.984. At the same time, the plant detection rate increased by 13 times compared to the model proposed earlier in the literature. Full article

The strategic allocation of advertising billboards has become a critical aspect of urban planning and resource management. While previous studies have explored site selection based on road network and population data, they have often overlooked the diminishing marginal returns of overlapping coverage and neglected to efficiently process large-scale urban datasets. To address these challenges, this study proposes two complementary optimization methods: an enhanced greedy algorithm based on geometric modeling and spatial acceleration techniques, and a reinforcement learning approach using Proximal Policy Optimization (PPO). The enhanced greedy algorithm incorporates population-weighted road coverage modeling, employs a geometric series to capture diminishing returns from overlapping coverage, and integrates spatial indexing and parallel computing to significantly improve scalability and solution quality in large urban networks. Meanwhile, the PPO-based method models billboard site selection as a sequential decision-making process in a dynamic environment, where agents adaptively learn optimal deployment strategies through reward signals, balancing coverage gains and redundancy penalties and effectively handling complex multi-step optimization tasks. Experiments conducted on Wuhan’s road network demonstrate that both methods effectively optimize population-weighted billboard coverage under budget constraints while enhancing spatial distribution balance. Quantitatively, the enhanced greedy algorithm improves coverage effectiveness by 18.6% compared to the baseline, while the PPO-based method further improves it by 4.3% with enhanced spatial equity. The proposed framework provides a robust and scalable decision-support tool for urban advertising infrastructure planning and resource allocation. Full article