Search Results (48)

Early detection of cotton diseases is critical for safeguarding crop yield and minimizing agrochemical usage. However, most state-of-the-art systems rely on multispectral or hyperspectral sensors, which are costly and inaccessible to smallholder farmers. This paper introduces CottoNet, a lightweight and efficient deep learning framework for detecting early-stage cotton diseases using only RGB images captured by unmanned aerial vehicles (UAVs). The proposed model integrates an EfficientNetV2-S backbone with a Dual-Attention Feature Pyramid Network (DA-FPN) and a novel Early Symptom Emphasis Module (ESEM) to enhance sensitivity to subtle visual cues such as chlorosis, minor lesions, and texture irregularities. A custom-labeled dataset was collected from cotton fields in Uzbekistan to evaluate the model under realistic agricultural conditions. CottoNet achieved a mean average precision (mAP@50) of 89.7%, an F1 score of 88.2%, and an early detection accuracy (EDA) of 91.5%, outperforming existing lightweight models while maintaining real-time inference speed on embedded devices. The results demonstrate that CottoNet offers a scalable, accurate, and field-ready solution for precision agriculture in resource-limited settings. Full article

Sound event detection tasks are rapidly advancing in the field of pattern recognition, and deep learning methods are particularly well suited for such tasks. One of the important directions in this field is to detect the sounds of emotional events around residential buildings in smart cities and quickly assess the situation for security purposes. This research presents a comprehensive study of an ensemble convolutional recurrent neural network (CRNN) model designed for sound event detection (SED) in residential and public safety contexts. The work focuses on extracting meaningful features from audio signals using image-based representation, such as Discrete Cosine Transform (DCT) spectrograms, Cocheagrams, and Mel spectrograms, to enhance robustness against noise and improve feature extraction. In collaboration with police officers, a two-hour dataset consisting of 112 clips related to four classes of emotional sounds, such as harassment, quarrels, screams, and breaking sounds, was prepared. In addition to the crowdsourced dataset, publicly available datasets were used to broaden the study’s applicability. Our dataset contains 5055 audio files of different lengths totaling 14.14 h and strongly labeled data. The dataset consists of 13 separate sound categories. The proposed CRNN model integrates spatial and temporal feature extraction by processing these spectrograms through convolution and bi-directional gated recurrent unit (GRU) layers. An ensemble approach combines predictions from three models, achieving F1 scores of 71.5% for segment-based metrics and 46% for event-based metrics. The results demonstrate the model’s effectiveness in detecting sound events under noisy conditions, even with a small, unbalanced dataset. This research highlights the potential of the model for real-time audio surveillance systems using mini-computers, offering cost-effective and accurate solutions for maintaining public order. Full article

This study presents a comparative analysis of the computed tomographic (CT), radiographic, and ultrasonographic (US) characteristics of gastrointestinal foreign bodies, including bezoars, in dogs and cats, and evaluates their association with complications and clinical outcomes. A total of 33 cases (26 dogs, 7 cats) with surgically or endoscopically confirmed foreign bodies were reviewed, classified as bezoars (n = 15) or distinct foreign bodies (n = 18). CT features such as attenuation values, transition zones, and proximal-to-distal small intestinal diameter ratios were compared. Bezoars typically appeared as intraluminal masses with mottled gas patterns and indistinct boundaries (33.3% vs. 94.4%, p < 0.001) and were associated with longer clinical signs (median 14 vs. 5.5 days, p = 0.013), more frequent transition zones (92.3% vs. 41.7%, p = 0.011), and a greater diameter ratio (2.9 vs. 1.25, p = 0.012) across the transition zone. Radiographic and US evaluations were available in six bezoar cases; only one radiograph (17%) detected the bezoar, while US showed acoustic shadowing in four cases (67%). Six patients (18%) experienced adverse outcomes, with bowel wall ruptures significantly associated with poor prognosis (p < 0.001). These findings highlight the superior diagnostic performance of CT, particularly for bezoars, and emphasize the importance of identifying transition zones and bowel diameter ratios in assessing gastrointestinal foreign bodies and their associated risks. Early CT evaluation may thus facilitate timely intervention and improve clinical outcomes. Full article

Parking management systems play a crucial role in addressing parking shortages and operational challenges; however, high initial costs and infrastructure requirements often hinder their implementation. Edge computing offers a promising solution by reducing latency and network traffic, thus optimizing operational costs. Nonetheless, the limited computational resources of edge devices remain a significant challenge. This study developed a real-time vehicle occupancy detection system utilizing SSD-MobileNetv2 on edge devices to process video streams from multiple IP cameras. The system incorporates a dual-trigger mechanism, combining periodic triggers and parking space mask triggers, to optimize computational efficiency and resource usage while maintaining high accuracy and reliability. Experimental results demonstrated that the parking space mask trigger significantly reduced unnecessary AI model executions compared to periodic triggers, while the dual-trigger mechanism ensured consistent updates even under unstable network conditions. The SSD-MobileNetv2 model achieved a frame processing time of 0.32 s and maintained robust detection performance with an F1-score of 0.9848 during a four-month field validation. These findings validate the suitability of the system for real-time parking management in resource-constrained environments. Thus, the proposed smart parking system offers an economical, viable, and practical solution that can significantly contribute to developing smart cities. Full article

The Advanced Insect Detection Network (AIDN), which represents a significant advancement in the application of deep learning for ecological monitoring, is specifically designed to enhance the accuracy and efficiency of insect detection from unmanned aerial vehicle (UAV) imagery. Utilizing a novel architecture that incorporates advanced activation and normalization techniques, multi-scale feature fusion, and a custom-tailored loss function, the AIDN addresses the unique challenges posed by the small size, high mobility, and diverse backgrounds of insects in aerial images. In comprehensive testing against established detection models, the AIDN demonstrated superior performance, achieving 92% precision, 88% recall, an F1-score of 90%, and a mean Average Precision (mAP) score of 89%. These results signify a substantial improvement over traditional models such as YOLO v4, SSD, and Faster R-CNN, which typically show performance metrics approximately 10–15% lower across similar tests. The practical implications of AIDNs are profound, offering significant benefits for agricultural management and biodiversity conservation. By automating the detection and classification processes, the AIDN reduces the labor-intensive tasks of manual insect monitoring, enabling more frequent and accurate data collection. This improvement in data collection quality and frequency enhances decision making in pest management and ecological conservation, leading to more effective interventions and management strategies. The AIDN’s design and capabilities set a new standard in the field, promising scalable and effective solutions for the challenges of UAV-based monitoring. Its ongoing development is expected to integrate additional sensory data and real-time adaptive models to further enhance accuracy and applicability, ensuring its role as a transformative tool in ecological monitoring and environmental science. Full article

Objectives: This study aimed to predict the unknown etiology of fundus-obscuring vitreous hemorrhage (FOVH) based on preoperative conditions using machine learning (ML) and to identify key preoperative factors. Methods: Medical records of 223 eyes from 204 patients who underwent vitrectomy for FOVH of unknown etiology between January 2012 and July 2022 were retrospectively reviewed. Preoperative data, including demographic information, systemic disease, ophthalmic history, and retinal status of the unaffected eye, were collected. The postoperatively identified etiologies of FOVH were categorized into six groups: proliferative diabetic retinopathy (PDR), retinal vein occlusion (RVO) or rupture of retinal arterial macroaneurysm, neovascular age-related macular degeneration (nAMD), retinal tear, Terson syndrome, and other causes. Four ML algorithms were trained and evaluated using seven-fold cross-validation. Results: The ML algorithms achieved mean accuracies of 76.2% for artificial neural network, 74.5% for XG-Boost, 74.4% for LASSO logistic regression, and 68.5% for decision tree. Key predictive factors commonly selected by the ML algorithms included PDR in the fellow eye, underlying diabetes mellitus, subarachnoid hemorrhage, and a history of retinal tear, RVO, or nAMD in the affected eye. Conclusions: The unknown etiology of FOVH could be predicted preoperatively with considerable accuracy by ML algorithms. Previous ophthalmic conditions in the affected eye and the condition of the fellow eye were important variables for prediction. This approach might assist in determining appropriate treatment plans. Full article

This study proposes an eXtended Reality (XR) glasses-based walking assistance system to support independent and safe outdoor walking for visually impaired people. The system leverages the YOLOv8n deep learning model to recognize walkable areas, public transport facilities, and obstacles in real time and provide appropriate guidance to the user. The core components of the system are Xreal Light Smart Glasses and an Android-based smartphone, which are operated through a mobile application developed using the Unity game engine. The system divides the user’s field of vision into nine zones, assesses the level of danger in each zone, and guides the user along a safe walking path. The YOLOv8n model was trained to recognize sidewalks, pedestrian crossings, bus stops, subway exits, and various obstacles on a smartphone connected to XR glasses and demonstrated an average processing time of 583 ms and an average memory usage of 80 MB, making it suitable for real-time use. The experiments were conducted on a 3.3 km route around Bokjeong Station in South Korea and confirmed that the system works effectively in a variety of walking environments, but recognized the need to improve performance in low-light environments and further testing with visually impaired people. By proposing an innovative walking assistance system that combines XR technology and artificial intelligence, this study is expected to contribute to improving the independent mobility of visually impaired people. Future research will further validate the effectiveness of the system by integrating it with real-time public transport information and conducting extensive experiments with users with varying degrees of visual impairment. Full article

This study utilized a genome-wide association study (GWAS) to investigate the genetic variations linked to the risk of hepatitis B virus (HBV) reactivation in patients who have undergone liver transplantation (LT), aiming to enhance understanding and improve clinical outcomes. Genotyping performed on a selected patients from the Korean Organ Transplantation Registry (KOTRY) data using high-throughput platforms with the Axiom Korea Biobank array 1.1. The discovery cohort included 21 patients who experienced HBV reactivation (cases) and 888 patients without HBV reactivation (controls) following LT. The replication cohort consisted of 5 patients with HBV reactivation (cases) and 312 patients without HBV reactivation (controls) after LT. Additive logistic regression analysis was conducted using PLINK software ver 1.9, with adjustments for age and gender. The GWAS findings from the discovery cohort were validated using the replication cohort. The GWAS identified several single-nucleotide polymorphisms (SNPs) in the RGL1, CDCA7L, and AQP9 genes that were significantly linked to HBV reactivation after LT, with genome-wide significance thresholds set at p < 10⁻⁷. Down-regulation of RGL1 cDNAs was observed in primary duck hepatocytes infected with duck HBV. Overexpression of CDCA7L was found to promote hepatocellular carcinoma cell proliferation and colony formation, whereas knocking down CDCA7L inhibited these processes. Additionally, the absence of AQP9 triggered immune and inflammatory responses, leading to mild and scattered liver cell pyroptosis, accompanied by compensatory liver cell proliferation. This study provides critical insights into the genetic factors influencing HBV reactivation after LT, identifying significant associations with SNPs in RGL1, CDCA7L, and AQP9. These findings hold promise for developing predictive biomarkers and personalized management strategies to improve outcomes for HBV-infected LT recipients. Full article

Researchers have studied instances of power line technical failures, the significant rise in the energy loss index in the line connecting the distribution transformer and consumer meters, and the inability to control unauthorized line connections. New, innovative, and scientific approaches are required to address these issues while enhancing the reliability and efficiency of electricity supply. This study evaluates the reliability of Internet of Things (IoT)-aided remote monitoring systems specifically designed for a low-voltage overhead transmission line. Many methods of analysis and comparison have been employed to examine the reliability of wireless sensor devices used in real-time remote monitoring. A reliability model was developed to evaluate the reliability of the monitoring system in various situations. Based on the developed models, it was found that the reliability indicators of the proposed monitoring system were 98% in 1 month. In addition, it has been proven that the reliability of the system remains high even when an optional sensor in the network fails. This study investigates various IoT technologies, their integration into monitoring systems, and their effectiveness in enhancing the reliability and efficiency of electrical transmission infrastructure. The analysis includes data from field deployments, case studies, and simulations to assess performance metrics, such as accuracy, latency, and fault detection capabilities. Full article

As an alternative to flat architectures, clustering architectures are designed to minimize the total energy consumption of sensor networks. Nonetheless, sensor nodes experience increased energy consumption during data transmission, leading to a rapid depletion of energy levels as data are routed towards the base station. Although numerous strategies have been developed to address these challenges and enhance the energy efficiency of networks, the formulation of a clustering-based routing algorithm that achieves both high energy efficiency and increased packet transmission rate for large-scale sensor networks remains an NP-hard problem. Accordingly, the proposed work formulated an energy-efficient clustering mechanism using a chaotic genetic algorithm, and subsequently developed an energy-saving routing system using a bio-inspired grey wolf optimizer algorithm. The proposed chaotic genetic algorithm–grey wolf optimization (CGA-GWO) method is designed to minimize overall energy consumption by selecting energy-aware cluster heads and creating an optimal routing path to reach the base station. The simulation results demonstrate the enhanced functionality of the proposed system when associated with three more relevant systems, considering metrics such as the number of live nodes, average remaining energy level, packet delivery ratio, and overhead associated with cluster formation and routing. Full article

Monocular depth prediction research is essential for expanding meaning from 2D to 3D. Recent studies have focused on the application of a newly proposed encoder; however, the development within the self-supervised learning framework remains unexplored, an aspect critical for advancing foundational models of 3D semantic interpretation. Addressing the dynamic nature of encoder-based research, especially in performance evaluations for feature extraction and pre-trained models, this research proposes the switchable encoder learning framework (SELF). SELF enhances versatility by enabling the seamless integration of diverse encoders in a self-supervised learning context for depth prediction. This integration is realized through the direct transfer of feature information from the encoder and by standardizing the input structure of the decoder to accommodate various encoder architectures. Furthermore, the framework is extended and incorporated into an adaptable decoder for depth prediction and camera pose learning, employing standard loss functions. Comparative experiments with previous frameworks using the same encoder reveal that SELF achieves a 7% reduction in parameters while enhancing performance. Remarkably, substituting newly proposed algorithms in place of an encoder improves the outcomes as well as significantly decreases the number of parameters by 23%. The experimental findings highlight the ability of SELF to broaden depth factors, such as depth consistency. This framework facilitates the objective selection of algorithms as a backbone for extended research in monocular depth prediction. Full article

The demand for customer support call centers has surged across various sectors due to the pandemic. Yet, the constraints of round-the-clock human services and fluctuating wait times pose challenges in fully meeting customer needs. In response, there’s a growing need for automated customer service systems that can provide responses tailored to specific domains and in the native languages of customers, particularly in developing nations like Uzbekistan where call center usage is on the rise. Our system, “UzAssistant,” is designed to recognize user voices and accurately present customer issues in standardized Uzbek, as well as vocalize the responses to voice queries. It employs feature extraction and recurrent neural network (RNN)-based models for effective automatic speech recognition, achieving an impressive 96.4% accuracy in real-time tests with 56 participants. Additionally, the system incorporates a sentence similarity assessment method and a text-to-speech (TTS) synthesis feature specifically for the Uzbek language. The TTS component utilizes the WaveNet architecture to convert text into speech in Uzbek. Full article

In recent years, parking lot management systems have garnered significant research attention, particularly concerning the application of deep learning techniques. Numerous approaches have emerged for tackling parking lot occupancy challenges using deep learning models. This study contributes to the field by addressing a critical aspect of parking lot management systems: accurate vehicle occupancy determination in specific parking spaces. We propose an advanced solution by harnessing an optimized MobileNetV3 model with custom architectural enhancements, trained on the CNRPark-EXT and PKLOT datasets. The model processes individual parking space patches from real-time video feeds, providing occupancy classification for each patch, identifying occupied or available spaces. Our architectural modifications include the integration of a convolutional block attention mechanism in place of the native attention module and the adoption of blueprint separable convolutions instead of the traditional depth-wise separable convolutions. In terms of performance, our proposed model exhibits superior results when benchmarked against state-of-the-art methods. Achieving an exceptional area under the ROC curve (AUC) value of 0.99 for most experiments with the PKLot dataset, our enhanced MobileNetV3 showcases its exceptional discriminatory power in binary classification. Benchmarked against the CarNet and mAlexNet models, representative of previous state-of-the-art solutions, our proposed model showcases exceptional performance. During evaluations using the combined CNRPark-EXT and PKLot datasets, the proposed model attains an impressive average accuracy of 98.01%, while CarNet achieves 97.03%. Beyond achieving high accuracy and precision comparable to previous models, the proposed model exhibits promise for real-time applications. This work contributes to the advancement of parking lot occupancy detection by offering a robust and efficient solution with implications for urban mobility enhancement and resource optimization. Full article

Soil moisture (SM) is a crucial hydrologic factor that affects the global cycle of energy, carbon, and water, as well as plant growth and crop yield; therefore, an accurate estimate of SM is important for both the global environment and agriculture. Satellite-based SM data have been provided by the National Aeronautics and Space Administration (NASA)’s Soil Moisture Active Passive (SMAP) and the European Space Agency (ESA)’s Soil Moisture and Ocean Salinity (SMOS) satellite missions, but these data are based on passive microwave sensors, which have limited spatial resolution. Thus, detailed observations and analyses of the local distribution of SM are limited. The recent emergence of deep learning techniques, such as rectified linear unit (ReLU) and dropout, has produced effective solutions to complex problems. Deep neural networks (DNNs) have been used to accurately estimate hydrologic factors, such as SM and evapotranspiration, but studies of SM estimates derived from the joint use of DNN and high-resolution satellite data, such as Sentinel-1 and Sentinel-2, are lacking. In this study, we aim to estimate high-resolution SM at 30 m resolution, which is important for local-scale SM monitoring in croplands. We used a variety of input data, such as radar factors, optical factors, and vegetation indices, which can be extracted from Sentinel-1 and -2, terrain information (e.g., elevation), and crop information (e.g., cover type and month), and developed an integrated SM model across various crop surfaces by using these input data and DNN (which can learn the complexity and nonlinearity of the various data). The study was performed in the agricultural areas of Manitoba and Saskatchewan, Canada, and the in situ SM data for these areas were obtained from the Agriculture and Agri-Food Canada (AAFC) Real-time In Situ Soil Monitoring for Agriculture (RISMA) network. We conducted various experiments with several hyperparameters that affected the performance of the DNN-based model and ultimately obtained a high-performing SM model. The optimal SM model had a root-mean-square error (RMSE) of 0.0416 m³/m³ and a correlation coefficient (CC) of 0.9226. This model’s estimates showed better agreement with in situ SM than the SMAP 9 km SM. The accuracy of the model was high when the daily precipitation was zero or very low and also during the vegetation growth stage. However, its accuracy decreased when precipitation or the vitality of the vegetation were high. This suggests that precipitation affects surface erosion and water layer formation, and vegetation adds complexity to the SM estimate. Nevertheless, the distribution of SM estimated by our model generally reflected the local soil characteristics. This work will aid in drought and flood prevention and mitigation, and serve as a tool for assessing the potential growth of crops according to SM conditions. Full article