Dense Layer Research Articles

Experimental Study on the Flexural Performance of Geogrid-Reinforced Foamed Lightweight Soil

The flexural behavior of geogrid-reinforced foamed lightweight soil (GRFL soil) is investigated in this study using unconfined compressive and four-point bending tests. The effects of wet density and reinforcement layers on flexural performance are analyzed using load–displacement curves, damage patterns, load characteristics, unconfined compressive strength, and flexural strength. A variance study demonstrates that increasing the wet density significantly increases unconfined compressive strength. Bond stress mechanisms enable geogrid integration, efficiently reroute stresses internally, and greatly increase flexural strength. With a maximum unconfined compressive strength of 3.16 MPa and a peak flexural strength increase of 166%, this reinforcement increases both strength and ductility by changing the damage pattern from brittle to ductile. The principal load is initially supported by the foamed lightweight soil, and in later phases, geogrids take over load-bearing responsibilities. Additionally, the work correlates the ratio of unconfined compressive to flexural strength with wet density and informs the development of predictive models for unconfined compressive strength as a function of reinforcing layers and wet density.

Hydrophobic modification of cellulose nanofibers/bionic flower-like ZnO synergistically stabilized Pickering emulsion to enhance pesticide deposition.

Environmental issues arising from the low pesticide utilization rate make the development of environmentally friendly and low-cost pesticide carrier systems an urgent problem to be solved. Pickering emulsion systems have shown broad application prospects in pesticide delivery. In this study, dodecenyl succinic anhydride (DDSA) was used to hydrophobically modify cellulose nanofiber (D-CNF), and biomimetic flower-like zinc oxide (ZnO) particles were prepared by precipitation method at room temperature. Then the two functioned synergistically as stabilizers for the Pickering emulsion, which significantly diminished the reliance on conventional surfactants. Turpentine, an essential oil, was used as a solvent for the broad-spectrum herbicide 2,4-dichlorophenoxyacetic acid (2,4-D) and as a part of the oil phase in an emulsion, replacing traditional toxic solvents for pesticide dissolution in the construction of a Pickering emulsion system. ZnO/D-CNF formed a dense layer at the oil-water interface to prevent droplet aggregation. After the Pickering emulsion contacted the leaves, ZnO could be embedded and adhere to leaves due to its morphology, which effectively reduced the splash of the droplets. No matter what the inclination angle was, the droplets could adhere and diffuse on the leaves. This study presents a new idea for creating a simple, eco-friendly, and practical pesticide carrier system.

Enhancing Colorectal Polyps Detection using Transfer Learning on DICOM Metadata

Colorectal and Rectum Cancer (CRC) presents significant global health challenges, necessitating early detection and precise diagnosis to achieve effective treatment and better patient outcomes. Transfer learning techniques have shown considerable promise, especially in cancer detection. This study presents a CRC prevention system based on a fusion of a pre-trained VGG16 model with dense layers for metadata processing. Experiments were performed using the CT Colonography dataset from The Cancer Imaging Archive (TCIA), applying preprocessing and class weighting to address class imbalance. The system was evaluated using accuracy, loss, recall, precision, F1-score, and AUC. This study investigated the impact of integrating DICOM patient metadata to enhance the proposed CRC prevention system. The findings indicate that the proposed MetaVGGNet model outperformed the standard VGG16, achieving greater accuracy (82%) and a marginally lower loss. This successful application has the potential to enhance CRC diagnosis and treatment and underscores the importance of incorporating metadata into deep learning classification systems, offering avenues for more effective and dependable diagnostic tools in CRC management.

The role of shocks and the velocity gradient in the relative orientation of the magnetic field and dense gas clouds

Magnetic fields are known to exhibit different relative orientations with density structures in different density regimes. However, the physical mechanisms behind these relative orientations remain unclear. We investigate the role of the flow features on the relative orientation between the magnetic field and cold neutral medium (CNM) clouds, as well as that of molecular clouds (MCs) as a corollary. We performed three- and two-dimensional (3D+2D) magnetohydrodynamic (MHD) simulations of warm gas streams in the thermally bistable atomic interstellar medium (ISM) colliding with velocities of the order of the velocity dispersion in the ISM to form CNM clouds. In these simulations, we followed the evolution of magnetic field lines to identify and elucidate the physical processes behind their evolution. The collision produces a fast MHD shock, as well as a condensation front roughly one cooling length behind it, on each side of the collision front. A compressive, decelerating velocity field arises between the shock and the condensation fronts, and a cold dense layer forms behind the condensation front. The magnetic field lines, initially oriented parallel to the flow direction, are perturbed by the fast MHD shock, across which the magnetic field fluctuations parallel to the shock front are amplified. The downstream perturbations of the magnetic field lines are further amplified by the compressive downstream velocity gradient between the shock and the condensation front caused by the settlement of the gas onto the dense layer. This process causes the magnetic field to become progressively aligned with the dense layer, leading to the formation of a shear flow around it, due to the field's backreaction on the flow. By extension, we suggest that a tidal stretching velocity gradient, such as that produced in gas infalling into a self-gravitating structure, must straighten the field lines along the accretion flow, orienting them perpendicular to the density structures. We also find that the initially super-Alfv'enic upstream flow becomes trans-Alfv'enic between the shock and the condensation front, and then sub-Alfv'enic inside the condensation. Finally, in 2D simulations with a curved collision front, the presence of the magnetic field inhibits the generation of turbulence by the shear around the dense layer. Our results provide a feasible physical mechanism for orienting the magnetic field parallel to CNM clouds through the action of fast MHD shocks and compressive velocity fields.

The evaluation of the macular and optic nerve head microvasculature in amblyopic and non-amblyopic patients with constant exotropia in comparison to healthy controls.

to evaluate the retinal microvascular changes in strabismic amblyopia and compare them to the strabismic cases without amblyopia and controls. Cross-sectional study PARTICIPANTS: : Three groups of patients, including 25 amblyopic cases with exotropia (XT), 25 constant XT cases without amblyopia, and 25 healthy subjects as controls, are included in the study. Both eyes of the included participants are evaluated. All the patients underwent a thorough ophthalmic examination, BCVA assessment, and macular and optic nerve head OCTA with the AngioVue™ Imaging System. The vascular density (VD) of various layers (Deep and superficial capillary plexus (DCP and SCP) in different regions of the macula, optic nerve head, and FAZ area were evaluated as the target variables. 42 FEMALES AND 33 MALES WITH A MEAN AGE OF 32.2 (95%CI: 29.8-34.6) entered our study. The median exotropia in both groups was 35 PD. The parafoveal DCP VD was significantly lower in both eyes of amblyopic cases than in controls (53.56 and 52.31 vs. 56.13%, p:0.021, <0.001). The parafoveal SCP VD was only lower in the dominant eye of the amblyopic cases than in the controls (50.53 vs. 53.54%, p:0.002). The foveal and peripapillary VD showed no significant difference between the groups. The FAZ area was significantly larger in amblyopic eyes than the fellow eyes (0.250 vs. 0.194 mm2, p:0.025) and correlated with worse BCVA (Correlation Coefficient: 0.358, p: 0.011). The amblyopia could be accompanied by a lower parafoveal DCP VD and a larger FAZ area in strabismic amblyopia.

A Fine-Tuned Adaptive Weight Deep Dense Meta Stacked Transfer Learning Model for Effective Cervical Cancer Prediction

Abstract In the digital world of remarkable technological advancements, the detection of cervical cancer at early stages is of important clinical significance as it can vastly improve the survival rate during treatment. Cervical cytopathology, often known as a Pap test is the frequently adopted screening method for cervical cancer. However, the test seems to be effective but investigation of images containing Pap smear with the help of a microscope is a difficult as well as laborious exercise. The procedure for the same demands an expert in the area and is often time-consuming. The serious pitfalls in subjective clinical evaluation evoke the need of developing an automated system for more reliable cervical cancer diagnosis. Therefore, the goal of this study primarily focuses on designing a Deep learning model to process the Pap smear images and correctly classify the cervical cells. For this purpose, firstly, a publically available dataset namely SIPaKMeD is utilized. Then, different data pre-processing methods are applied to intensify the data quality for effective analysis. Next, a novel stacking model is proposed that leverages a Support Vector Classifier (SVC) as a Meta model over a combination of different Transfer Learning Models including VGG16, ResNet101, InceptionV3, Xception, DenseNet169, and Inception ResNet. Furthermore, the dense layers are added to tune the underlying base transfer learning models to learn fine-tuned adaptive weights. The results obtained from experimental evaluation demonstrate the efficacy of the proposed stacking model by yielding the highest accuracy rate of 95.66% in comparison to other employed methods and existing state-of-the-art techniques.

Estimation of Carbon Density in Different Urban Green Spaces: Taking the Beijing Main District as an Example

Urban green spaces (UGS) are crucial urban elements that serve as direct carbon sequestration and contribute to indirect carbon emission reduction. Accurately calculating the carbon density of urban green spaces allows for scientific planning and design, thereby advancing efforts toward achieving carbon neutrality. This study has developed a workflow for estimating carbon density in urban green spaces through point cloud measurements and model simulations, using the UGS in the Beijing Main District as a case study. From the sample level, a calculation methodology was constructed based on the point cloud technology-model simulation method, which can obtain the carbon density at the plant level and the sample level. At the UGS level, remote sensing inversion was utilized to map the carbon density of urban green spaces. Ultimately, the research calculated and compared carbon density at different scales, including the carbon density of individual plants, the carbon density of sample plots, and the carbon density of various types of urban green spaces. It was found that the carbon density of trees in UGS was 9.87 kg/m2, while those of shrubs and herbaceous plants were 13.20 kg/m2 and 0.11 kg/m2. In urban green spaces, the carbon densities of the tree and herb layers were slightly lower than those in natural ecosystems, whereas the carbon density of the shrub layer was significantly higher. This highlights the substantial potential and value of shrubs in carbon sequestration and carbon storage. The average carbon density of all UGS types was 9.76 kg/m2, with the following descending order: Neighborhood Parks (10.31 kg/m2) &gt; Attached Green Spaces (7.22 kg/m2) &gt; Regional Parks (5.75 kg/m2). Based on these findings, the study proposed optimization strategies for different UGS types, focusing on high carbon-density plant community optimization. The goal is to provide a theoretical foundation for carbon storage calculations and plant arrangements in future UGS construction.

“Advancing Ozone Forecasting with LSTM Neural Networks: A Study on Time Window Variations and Performance Metrics”

ABSTRACT In the face of growing global concern about air pollution and its adverse impacts on health and well-being, air quality forecasting has become a major focus of research. This study investigates the application of Long Short-Term Memory (LSTM) neural network models for the prediction of ozone concentrations in different time windows (7, 15 and 30 days). The LSTM models were configured with two LSTM layers and two dense layers, varying the number of neurons between 50 and 200. Predictor variables included ‘Month,’ CO_ppb,’ ‘SO2_ugm3,’ ‘temperature,’ ‘humidity’ and ‘PM2.5_ugm3,’ while the output variable was ‘O3_ppb.’ The results showed that the LSTM model with 200 LSTM units and 50 dense units had the best performance, presenting a mean absolute error (MAE) of 1.49 ppb, an RMSE of 1.91 ppb and a coefficient of determination (R2) of 0.94 for 7-day forecasts. These results outperformed traditional models such as ARMA and linear regression, highlighting the effectiveness of LSTM in capturing complex patterns in ozone time series. The research suggests that future investigations could explore hybrid techniques to further improve the accuracy of predictions.

Targeted cattle grazing for shrub control in woody‐encroached oak savannas

Woody encroachment threatens savanna ecosystems around the world. In the midwestern United States, remnant oak savannas persist in a woody‐encroached state due to the removal of grazing and fire disturbances from the landscape. Primary oak savanna restoration methods, such as thinning and fire, can promote a dense shrub layer. Therefore, it is necessary to explore other tools to control woody species, such as targeted grazing. We conducted an experiment to evaluate the effectiveness of targeted cattle grazing as a shrub control method and its impacts on the herbaceous community. We used high‐density, short‐duration grazing to target shrubs and minimize negative non‐target impacts. We found that this type of cattle management reduced shrub density over the short term, but there was substantial resprout after 1 year of no grazing. Targeted grazing lowered herbaceous cover and shifted species composition but did not affect herbaceous species richness or diversity. We saw no increase in non‐native cover during the study, though we documented isolated occurrences of new non‐native species in the grazing paddocks. Overall, our study points to the potential of targeted cattle grazing as a viable tool to help restore the understory of woody‐encroached ecosystems. Successfully meeting vegetation objectives requires a grazing manager with knowledge of the site, the target species, and livestock care and nutrition, as well as a commitment to adaptive management.

The evaluation of giant-cell arteritis (temporal arteritis) cases with optical coherence tomography angiography (OCT-A)

To make measurements using optical coherence tomography angiography (OCT-A) in inactive giant cell arteritis (GCA) cases who have previously had GCA and have been treated and to compare the obtained data with healthy volunteers. In this observational case-control study, 18 eyes of 18 GCA cases previously diagnosed, treated with anterior arteritic ischemic optic neuropathy (AAION). 22 eyes of 22 ophthalmically healthy volunteers were included in the study. After external ophthalmic examinations of all participants were performed, their measurements were made with serial OCT-A. Superficial capillary plexus (SCP), deep capillary plexus (DCP), foveal avascular zone (FAZ), area covering 300 degrees around the fovea (FD-300), choriocapillaris (CC), retinal nerve fiber layer (RNFL), cup/disc (C/D) ratio and optic disc vessel densities (OD-VD) were evaluated. p&lt;0.05 was considered significant. There was no difference between the two groups in terms of age, gender and shooting quality. Whole-SCP, SCP-foveal, SCP-parafoveal and SCP-perifoveal VD values were lower in the patient group. Whole-DCP, DCPparafoveal and DCP-perifoveal VD values were also low in the patient group. FAZ areas were similar between groups, but the FD-300 VD was different. Whole-OD VD and inside-OD VD were significantly lower in the patient group. Peripapillary-OD VD and RNFL values were similar. The C/D ratio was higher in the patient group. The effect on the microvascular process was significant in OCT-A. This suggested that even if the ischemic process still continues and there is no active inflammation, microvascular structures may continue to be affected.

Quantitative analysis of peripapillary RNFL capillary density and thickness in patients with mild nonproliferative diabetic retinopathy

PurposeOptical coherence tomography angiography was used to compare the changes of capillary density and thickness of the peripapillary retinal nerve fiber layer in patients with diabetes mellitus (DM) without diabetic retinopathy (DR) and patients with mild nonproliferative diabetic retinopathy (NPDR).MethodsIn this prospective cross-sectional study, 62, 85, and 75 eyes of normal control group (NC), DM group, and NPDR group were included, respectively. All subjects received an optic-disc-centered 6 × 6-mm fundus scan. The annular region outside the optic disc boundary was divided into eight sectors: nasal superior, nasal inferior, inferior nasal (IN), inferior temporal (IT), temporal superior, temporal inferior, superior nasal, and superior temporal (ST). The average retinal nerve fiber layer thickness, average capillary density, and changes in thickness and capillary density of each sector in the annular region were calculated.ResultsCompared with the NC group, the average capillary density of pRNFL in patients with DM and NPDR groups decreased significantly (p &amp;lt; 0.001), and there were significant differences in capillary density in all sectors except the IN sector in the DM group. There was no significant difference in the average thickness of pRNFL between the groups. Only the thickness of ST and IT in DM group and NPDR group, respectively, were significantly lower than that in the control group. In addition, the ROC curve of pRNFL average capillary density has high sensitivity and specificity in distinguishing DR from healthy eyes (AUC = 0.852).ConclusionThe pRNFL average capillary density in diabetic patients without DR and patients with mild NPDR, respectively, were significantly lower than that in healthy controls, especially in the NPDR group.

Laminated Two-Terminal All-Perovskite Tandem Solar Cells with Transparent Conductive Adhesives.

Established sequential deposition of multilayer two-terminal (2T) all-perovskite tandem solar cells possesses challenges for fabrication and limits the choice of materials and device architecture. In response, this work represents a lamination process based on a transparent and conductive adhesive that interconnects the wide-bandgap (WBG) perovskite top solar cell and the narrow-bandgap (NBG) perovskite bottom solar cell in a monolithic 2T all-perovskite tandem solar cell. The transparent conductive adhesive (TCA) layer combines Ag-coated poly(methyl methacrylate) microspheres with an optical adhesive. The TCA is employed as a recombination junction, achieving self-encapsulation with high transparency and good conductivity. A high vertical electrical conductivity is realized by optimizing the distribution of microspheres using a novel solidification strategy that employs UV curing and drying at 85 °C at low pressure. In addition, uniform and dense bilayer hole transport layers are realized by vapor-phase evaporation, which facilitates the application of the TCA and achieves pinhole-free buried interfaces in both the WBG perovskite top solar cell and the NBG perovskite bottom solar cell. Using these two strategies, losses in fill factor (FF) and open-circuit voltage (VOC) of the 2T all-perovskite tandem solar cell are reduced, achieving a respectable power conversion efficiency up to 18.2% for the lamination of the all-perovskite tandem solar cell. The laminated tandem solar cell retains ∼93% of initial efficiency after exposure in ambient air (30-70 RH% and 20-35 °C) for ∼30 days.

Non-Linear Hyperelastic Model Analysis and Numerical Validation of 3D Printed PLA+ Material Incorporating Various Infill Densities

Additive manufacturing (AM) or 3D printing technology creates a tangible object by adding successive layers of materials. Nowadays, 3D printing is used for developing both metal and non-metal products. In the advancement of 3D printing technology, material specimen design, modification, and testing become very simple, especially for non-metal materials, such as hyperelastic, thermoplastic, or rubber-like materials. However, proper material modeling and validation are required for the analysis of these types of materials. In this study, 3D printed poly lactic acid (PLA+) material behavior is analyzed numerically for validation in the counterpart of experimental analysis to evaluate their behavior in both cases. The specimen was designed in SolidWorks by following ASTM D638 dimension standards with proper infill densities and raster angle or infill orientation angle. These infill layer densities and angles of orientation play an important role in the mechanical behavior of the specimen. This paper aims to present a numerical validation of five infill densities (20%, 40%, 60%, 80%, and 100%) for a ±45-degree infill angle orientation by incorporating a nonlinear hyperelastic model. Results indicate that infill densities affect the mechanical behavior of PLA+ material. The result also suggested that neo-Hookean and Mooney–Rivlin are the best-fitted hyperelastic material models for these five separate linear infill densities. However, neo-Hookean is easier to analyze, as it has only one parameter and a new equation is developed in this study for determining the parameter for different infill densities.

Agarista hispidula var. tomentosa (Ericaceae), a new variety from Minas Gerais, Brazil

Agarista (Ericaceae) is a Neotropical genus of subshrubs, shrubs and trees in Ericaceae, highly associated with tropical mountainous environments, especially the Campos Rupestres of the Cerrado domain in the Espinhaço Range, southeast Brazil. The genus comprises 35 species and six varieties. Minas Gerais state hosts a high richness of the genus, comprising 16 species, eight of them endemic. Here we indicate two partially geographically overlapping and morphologically characterized varieties, separable primarily by the indumentum on the abaxial leaf surface. The new variety Agarista hispidula var. tomentosa from Minas Gerais state is described here, and it is readily distinguished from Agarista hispidula var. hispidula by its dense layer of unicellular hairs on the abaxial leaf surface covering the abaxial epidermis. This character state is also found in three other species and two varieties of Agarista, and it may function in avoiding water loss. A detailed description and photograph plates are provided for the new variety. A key and a distribution map to the varieties of A. hispidula is also provided.

Lime and marble deposits mapping and estimation through deep neural layers-random forest merger and remote sensing imagery

Abstract Lime and marble are sedimentary rocks (also called carbonated rocks), mainly composed of calcite and dolomite as major ingredients. Calcite (CaCo3) and dolomite (MgCo3) offer excellent reflection in visible and short-wave infrared bands (0.4 to 2.5 µm), which distinguish them from other rock types and makes them a solid application area for multispectral sentinel-2 sensors. In this study, monumental deposits of carbonated rocks of the intended study region are mapped and estimated using a novel Deep Neural Layers-Random Forest merger model and Sentinel-2 imagery. We used ArcGIS 10.5 to prepare a high-quality classified map of the study regions, which can be used further to effectively and efficiently reach out and extract those deposits. The proposed model, trained on enough purified data containing 131,750 pixels of size 10 m, mapped the deposits with 0.95 accuracy (kappa coefficient 0.94), which is 70 % of the total mineral deposits of the study regions. The proposed model is the improved version of a prior experimented model consisting of convolution layers and dense layers only, whose accuracy was 0.93.&amp;#xD;

High‐Efficiency PbS Quantum Dots Infrared Solar Cells via Numerical Simulation and Experimental Optimization

AbstractLow‐bandgap lead sulfide quantum dots (PbS QDs) can efficiently harness the infrared (IR) light in the solar spectrum beyond 1100 nm, showing great application potential in the bottom subcells of tandem solar cells. However, achieving further efficiency improvements in PbS QDs IR solar cells still faces many challenges. In this work, the effects of the absorber layer thickness, the carrier mobility in the absorber layer, the defect density in the absorber layer and at the absorber/electron transfer layer (ETL) interface, and the doping density of the ETL and hole transfer layer (HTL) on the performance of PbS QDs (≈0.95 eV) IR solar cells are systematically investigated through SCAPS‐1D simulation. A theoretical efficiency of 16.95% and 2.15% is calculated for PbS QDs IR solar cells under AM 1.5 and 1100 nm‐filtered illumination, respectively. Based on the simulation results, the corresponding PbS QDs IR solar cells are fabricated with an efficiency of 11.53% under AM 1.5 illumination, a remarkable 1100 nm‐filtered efficiency of 1.30%, and a high external quantum efficiency of 70.50% at 1290 nm. Hence, these findings will accelerate the optimization of the performance of PbS QDs IR solar cells approaching their theoretical efficiency limit.

Tailoring selenization dynamics: How heating rate manipulates nucleation and growth boosts efficiency in kesterite solar cells.

Kesterite Cu2ZnSn(S,Se)4 (CZTSSe) has emerged as a promising photovoltaic material due to its low cost and high stability. The CZTSSe film for high-performance solar cells can be obtained by annealing the deposited CZTS precursor films with selenium (a process known as selenization). The design of the selenization process significantly affects the quality of the absorber layer. In this work, we systematically investigate the impact of heating rate on the selenization kinetics and the microstructural characteristics of the films using a two-step selenization method. The results indicate that a slow heating rate promotes surface crystallization, resulting in a thick and dense layer of large grains at the film surface that impedes the diffusion of Se vapor. Conversely, a rapid heating rate enhances the diffusion of Se into the interior of the film, synthesizing more low-melting-point intermediate compounds that facilitate grain growth and reduce the thickness of fine grains at the film bottom. Ultimately, a CZTSSe solar cell with an efficiency of 10.17% was fabricated at a heating rate of 200 °C/min. This research deepens the understanding of thin film growth mechanisms and advances the development of high-performance solar cells.

Electrostatic and Electronic Effects on Doped Nickel Oxide Nanofilms for Water Oxidation.

An ideal water-splitting electrocatalyst is inexpensive, abundant, highly active, stable, selective, and durable. The anodic oxygen evolution reaction (OER) is the main bottleneck for H2 production with a complex and not fully resolved mechanism, slow kinetics, and high overpotential. Nickel oxide-based catalysts (NiOx) are highly active and cheaper than precious metal catalysts. However, rigorous catalyst tests and DFT calculations are still needed to rationally optimize NiOx catalysts. In this work, we combine plasma-enhanced atomic layer deposition (PE-ALD) and density functional theory (DFT) to address the role of dopants in promoting NiOx OER activity. Ultrathin films of NiOx doped with Zn2+, Al3+, and Sn4+ presented improved intrinsic activity, stability, and durability for the OER. The results show a low to high catalytic performance of ZnNiOx < NiOx < AlNiOx < SnNiOx, which we attribute to an increase in the concentration of valence band (VB) holes combined with conduction band (CB) electron conductivity, characterized by electrochemical impedance spectroscopy (EIS). The influence of doping on the electronic structure and catalytic activity was investigated using advanced characterization techniques and density functional theory (DFT) calculations (PEB0/pob-TZVP). DFT complements the experimental results, showing that the dopant charge states and orbital hybridization enhance the OER by improving the charge carrier concentration and mobility, thus allowing optimal binding energies and charge dynamics and delocalization. Our findings demonstrate the potential of PE-ALD-doped nanofilms NiOx and DFT to rationally design and develop catalysts for sustainable energy applications.

GLAC-Unet: Global-Local Active Contour Loss with an Efficient U-Shaped Architecture for Multiclass Medical Image Segmentation.

The field of medical image segmentation powered by deep learning has recently received substantial attention, with a significant focus on developing novel architectures and designing effective loss functions. Traditional loss functions, such as Dice loss and Cross-Entropy loss, predominantly rely on global metrics to compare predictions with labels. However, these global measures often struggle to address challenges such as occlusion and nonuni-form intensity. To overcome these issues, in this study, we propose a novel loss function, termed Global-Local Active Contour (GLAC) loss, which integrates both global and local image features, reformulated within the Mumford-Shah framework and extended for multiclass segmentation. This approach enables the neural network model to be trained end-to-end while simultaneously segmenting multiple classes. In addition to this, we enhance the U-Net architecture by incorporating Dense Layers, Convolutional Block Attention Modules, and DropBlock. These improvements enable the model to more effectively combine contextual information across layers, capture richer semantic details, and mitigate overfitting, resulting in more precise segmentation outcomes. We validate our proposed method, namely GLAC-Unet, which utilizes the GLAC loss in conjunction with our modified U-shaped architecture, on three biomedical segmentation datasets that span a range of modalities, including two-dimensional and three-dimensional images, such as dermoscopy, cardiac magnetic resonance imaging, and brain magnetic resonance imaging. Extensive experiments demonstrate the promising performance of our approach, achieving a Dice score (DSC) of 0.9125 on the ISIC-2018 dataset, 0.9260 on the Automated Cardiac Diagnosis Challenge (ACDC) 2017, and 0.927 on the Infant Brain MRI Segmentation Challenge 2019. Furthermore, statistical significance testing with p-values consistently smaller than 0.05 on the ISIC-2018 and ACDC datasets confirms the superior performance of the proposed method compared to other state-of-the-art models. These results highlight the robustness and effectiveness of our multiclass segmentation technique, underscoring its potential for biomedical image analysis. Our code will be made available at https://github.com/minhnhattrinh312/Active-Contour-Loss-based-on-Global-and-Local-Intensity.

Applying convolutional neural networks for mustard variety recognition

The aim of this study was to develop and apply a Convolutional Neural Network (CNN) model to recognize and classify white mustard (Sinapis Alba L.) varieties, addressing the complex task of discriminating among 57 varieties. Utilizing a one-dimensional CNN model, the research focused on multivariate analysis based on a set of 15 traits. The CNN architecture included convolutional layers, batch normalization, pooling, flattening, dropout, and dense layers. The model demonstrated effectiveness in classifying varieties, achieving high accuracy and providing valuable insights into potential new varieties. Subset division, a new approach, was applied. Evaluation metrics, including accuracy, F1 score, precision, and recall, were calculated for eight subsets, confirming the model's robust performance. While this study uses mustard as an illustrative example, the method is not limited to this crop and can be extended to other agricultural crops, with potential modifications depending on the specific traits relevant to each crop. The approach contributes to agricultural advancements, offering a reliable tool for breeders to assess variety distinctness and streamline the testing process. The model’s ability to detect unknown varieties further enhances its utility in agricultural research covering a comprehensive and impactful advancement in variety classification.