• KIPS Transactions on Software and Data Engineering
• /
• v.12 no.12
• /
• pp.505-518
• /
• 2023

3D point cloud semantic segmentation is a computer vision task that involves dividing the point cloud into different objects and regions by predicting the class label of each point. Existing 3D semantic segmentation models have some limitations in performing sufficient fusion of multi-modal features while ensuring both characteristics of 2D visual features extracted from RGB images and 3D geometric features extracted from point cloud. Therefore, in this paper, we propose MMCA-Net, a novel 3D semantic segmentation model using 2D-3D multi-modal features. The proposed model effectively fuses two heterogeneous 2D visual features and 3D geometric features by using an intermediate fusion strategy and a multi-modal cross attention-based fusion operation. Also, the proposed model extracts context-rich 3D geometric features from input point cloud consisting of irregularly distributed points by adopting PTv2 as 3D geometric encoder. In this paper, we conducted both quantitative and qualitative experiments with the benchmark dataset, ScanNetv2 in order to analyze the performance of the proposed model. In terms of the metric mIoU, the proposed model showed a 9.2% performance improvement over the PTv2 model using only 3D geometric features, and a 12.12% performance improvement over the MVPNet model using 2D-3D multi-modal features. As a result, we proved the effectiveness and usefulness of the proposed model.

• Journal of the Korea Society for Simulation
• /
• v.33 no.1
• /
• pp.1-17
• /
• 2024

Forest management is known to beneficially influence stand structure and wood production, yet quantitative understanding as well as an illustrative depiction of the effects of different management approaches on tree growth and stand dynamics are still scarce. Long-term management of beech forests must balance public interests with ecological aspects. Efficient forest management requires the reliable prediction of tree growth change. We aimed to develop a novel hybrid simulation approach, which realistically simulates short- as well as long-term effects of different forest management regimes commonly applied, but not limited, to German low mountain ranges, including near-natural forest management based on single-tree selection harvesting. The model basically consists of three modules for (a) natural seedling regeneration, (b) mortality adjustment, and (c) tree growth simulation. In our approach, an existing validated growth model was used to calculate single year tree growth, and expanded on by including in a newly developed simulation process using calibrated modules based on practical experience in forest management and advice from the local forest. We included the following different beech forest-management scenarios that are representative for German low mountain ranges to our simulation tool: (1) plantation, (2) continuous cover forestry, and (3) reserved forest. The simulation results show a robust consistency with expert knowledge as well as a great comparability with mid-term monitoring data, indicating a strong model performance. We successfully developed a hybrid simulation that realistically reflects different management strategies and tree growth in low mountain range. This study represents a basis for a new model calibration method, which has translational potential for further studies to develop reliable tailor-made models adjusted to local situations in beech forest management.

• Korean Journal of Radiology
• /
• v.23 no.12
• /
• pp.1269-1280
• /
• 2022

Objective: This study aimed to evaluate the usefulness of quantitative indices obtained from deep learning analysis of gadoxetic acid-enhanced hepatobiliary phase (HBP) MRI and their longitudinal changes in predicting decompensation and death in patients with advanced chronic liver disease (ACLD). Materials and Methods: We included patients who underwent baseline and 1-year follow-up MRI from a prospective cohort that underwent gadoxetic acid-enhanced MRI for hepatocellular carcinoma surveillance between November 2011 and August 2012 at a tertiary medical center. Baseline liver condition was categorized as non-ACLD, compensated ACLD, and decompensated ACLD. The liver-to-spleen signal intensity ratio (LS-SIR) and liver-to-spleen volume ratio (LS-VR) were automatically measured on the HBP images using a deep learning algorithm, and their percentage changes at the 1-year follow-up (ΔLS-SIR and ΔLS-VR) were calculated. The associations of the MRI indices with hepatic decompensation and a composite endpoint of liver-related death or transplantation were evaluated using a competing risk analysis with multivariable Fine and Gray regression models, including baseline parameters alone and both baseline and follow-up parameters. Results: Our study included 280 patients (153 male; mean age ± standard deviation, 57 ± 7.95 years) with non-ACLD, compensated ACLD, and decompensated ACLD in 32, 186, and 62 patients, respectively. Patients were followed for 11-117 months (median, 104 months). In patients with compensated ACLD, baseline LS-SIR (sub-distribution hazard ratio [sHR], 0.81; p = 0.034) and LS-VR (sHR, 0.71; p = 0.01) were independently associated with hepatic decompensation. The ΔLS-VR (sHR, 0.54; p = 0.002) was predictive of hepatic decompensation after adjusting for baseline variables. ΔLS-VR was an independent predictor of liver-related death or transplantation in patients with compensated ACLD (sHR, 0.46; p = 0.026) and decompensated ACLD (sHR, 0.61; p = 0.023). Conclusion: MRI indices automatically derived from the deep learning analysis of gadoxetic acid-enhanced HBP MRI can be used as prognostic markers in patients with ACLD.

• Journal of Korea Society of Industrial Information Systems
• /
• v.29 no.1
• /
• pp.135-144
• /
• 2024

In this study, we propose an automatic classification model for quantitative multidimensional analysis based on facet theory to understand public opinions and demands on major issues through big data analysis. Civil complaints, as a form of public feedback, are generated by various individuals on multiple topics repeatedly and continuously in real-time, which can be challenging for officials to read and analyze efficiently. Specifically, our research introduces a new classification framework that utilizes facet theory and political analysis models to analyze the characteristics of citizen complaints and apply them to the policy-making process. Furthermore, to reduce administrative tasks related to complaint analysis and processing and to facilitate citizen policy participation, we employ deep learning to automatically extract and classify attributes based on the facet analysis framework. The results of this study are expected to provide important insights into understanding and analyzing the characteristics of big data related to citizen complaints, which can pave the way for future research in various fields beyond the public sector, such as education, industry, and healthcare, for quantifying unstructured data and utilizing multidimensional analysis. In practical terms, improving the processing system for large-scale electronic complaints and automation through deep learning can enhance the efficiency and responsiveness of complaint handling, and this approach can also be applied to text data processing in other fields.

• Land and Housing Review
• /
• v.15 no.1
• /
• pp.167-177
• /
• 2024

This study aimed to assess the efficacy of an adsorption process in removing organic matter and micropollutant residuals. After a full-scale water circulation system, the adsorption process was considered a post-treatment step. The system, treating anthropogenically impacted surface waters, comprises a hydro-cyclone, coagulation, flocculation, and dissolved air flotation unit. While the system generally maintained stable and satisfactory effluent quality standards over months, it did not meet the highest standard for organic matter (as determined by chemical oxygen demands). Adsorption experiments utilized two granular activated carbon types, GAC 830 and GCN 830, derived from coal and coconut-shell feedstocks, respectively. The assessment encompassed organic materials along with two notable micropollutants: acetaminophen (APAP) and acid orange 7 (AO7). Adsorption kinetics and isotherm experiments were conducted to determine adsorption rates and maximum adsorption amounts. The quantitative findings derived from pseudo-second-order kinetics and Langmuir isotherm models suggest the effectiveness of the adsorption process. The findings of this study propose the potential of employing the adsorption process as a post-treatment to enhance the treatment of contaminants that are not satisfactorily treated by conventional water circulation systems. This enhancement is crucial for ensuring the sustainability of urban water cycles.

• Journal of the Korea Society of Computer and Information
• /
• v.29 no.5
• /
• pp.21-29
• /
• 2024

Reducing the radiation dose during CT scanning can lower the risk of radiation exposure, but not only does the image resolution significantly deteriorate, but the effectiveness of diagnosis is reduced due to the generation of noise. Therefore, noise removal from CT images is a very important and essential processing process in the image restoration. Until now, there are limitations in removing only the noise by separating the noise and the original signal in the image area. In this paper, we aim to effectively remove noise from CT images using the wavelet transform-based GAN model, that is, the WT-GAN model in the frequency domain. The GAN model used here generates images with noise removed through a U-Net structured generator and a PatchGAN structured discriminator. To evaluate the performance of the WT-GAN model proposed in this paper, experiments were conducted on CT images damaged by various noises, namely Gaussian noise, Poisson noise, and speckle noise. As a result of the performance experiment, the WT-GAN model is better than the traditional filter, that is, the BM3D filter, as well as the existing deep learning models, such as DnCNN, CDAE model, and U-Net GAN model, in qualitative and quantitative measures, that is, PSNR (Peak Signal-to-Noise Ratio) and SSIM (Structural Similarity Index Measure) showed excellent results.

• Tunnel and Underground Space
• /
• v.34 no.3
• /
• pp.208-217
• /
• 2024

There is so much research effort for developing and implementing deep learning-based surveillance systems to manage health and safety issues in construction sites. Especially, the development of deep learning-based object detection in various environmental changes has been progressing because those affect decreasing searching performance of the model. Among the various environmental variables, the accuracy of the object detection model is significantly dropped under low illuminance, and consistent object detection accuracy cannot be secured even the model is trained using low-light images. Accordingly, there is a need of low-light enhancement to keep the performance under low illuminance. Therefore, this paper conducts a comparative study of various deep learning-based low-light image enhancement models (GLADNet, KinD, LLFlow, Zero-DCE) using the acquired construction site image data. The low-light enhanced image was visually verified, and it was quantitatively analyzed by adopting image quality evaluation metrics such as PSNR, SSIM, Delta-E. As a result of the experiment, the low-light image enhancement performance of GLADNet showed excellent results in quantitative and qualitative evaluation, and it was analyzed to be suitable as a low-light image enhancement model. If the low-light image enhancement technique is applied as an image preprocessing to the deep learning-based object detection model in the future, it is expected to secure consistent object detection performance in a low-light environment.

• The Korean Journal of Pesticide Science
• /
• v.14 no.4
• /
• pp.319-325
• /
• 2010

To understand the influences of the substituents ($R_1{\sim}R_4$) on insecticidal activity of N'-phenyl-N-methylformamidine analogues (1~22) against two spotted spider mite (Tetranychus urticae), comparative molecular field analysis (CoMFA) model and comparative molecular similarity indices analysis (CoMSIA) model as three dimensional quantitative structure-activity relationships (3D-QSARs) model were derived and discussed quantitatively. From the results, the correlativity and predictability ($r^2{_{cv.}}=0.575$ and $r^2{_{ncv.}}=0.945$) of the CoMFA 1 model were higher than those of the rest models. The the CoMFA 1 and CoMSIA 1 model with the sensitivity of the perturbation and the prediction produced ($d_q{^{2'}}/dr^2{_{yy}}=1.071{\sim}1.146$ & $q^2=0.545{\sim}0.626$) by a progressive scrambling analysis were not dependent on chance correlation. The insecticidal activities from the optimized CoMFA 1 model were depend upon the steric field (62.5%), electrostatic field (28.9%), and hydrophobic field (8.6%) of N'-phenyl-N-methylformamidine analogues. Therefore, the inhibitory activities with optimized CoMFA 1 model were dependent upon steric factor. From the contour maps of the optimized models, it is predicted that the structural distinctions that contribute to the insecticidal activity will be able to applied new potent insecticides design.

• Journal of Bio-Environment Control
• /
• v.24 no.3
• /
• pp.161-166
• /
• 2015

To determine the adequate levels of light intensity and $CO_2$ concentration for mango grown in greenhouses, quantitative measurements of photosynthetic rates at various leaf positions in the tree are required. The objective of this study was to develop two-variable leaf photosynthetic models of Irwin mango (Mangifera indica L. cv. Irwin) using light intensity and $CO_2$ concentration at different leaf positions. Leaf photosynthetic rates at different positions (top, middle, and bottom) were measured by a leaf photosynthesis analyzer at light intensities (0, 50, 100, 200, 300, 400, 600, and $800{\mu}mol{\cdot}m^{-2}{\cdot}s^{-1}$) with $CO_2$ concentrations (100, 400, 800, 1200, and $1600{\mu}mol{\cdot}mol^{-1}$). The two-variable model consisted of the two leaf photosynthetic models expressed as negative exponential functions for light intensity and $CO_2$ concentrations, respectively. The photosynthetic rates of top leaves were saturated at a light intensity of $400{\mu}mol{\cdot}^{-2}{\cdot}s^{-1}$, while those of middle and bottom leaves saturated at $200{\mu}mol{\cdot}^{-2}{\cdot}s^{-1}$. The leaf photosynthetic rates did not reach the saturation point at a $CO_2$ concentration of $1600imolmol^{-1}$. In validation of the model, the estimated photosynthetic rates at top and bottom leaves showed better agreements with the measured ones than the middle leaves. It is expected that the optimal conditions of light intensity and $CO_2$ concentration can be determined for maximizing photosynthetic rates of Irwin mango grown in greenhouses by using the two-variable model.

• Korean Journal of Soil Science and Fertilizer
• /
• v.43 no.4
• /
• pp.415-423
• /
• 2010

Pore network models are useful tools to investigate soil pore geometry. These models provide quantitative information of pore geometry from 3D images. This study presents a pore network model to quantify pore structure and hydraulic characteristics. The objectives of this work were to apply the pore network model to characterize pore structure from large images to quantify pore structure, calculate water retention and hydraulic conductivity properties from a three dimensional soil image, and to combine measured hydraulic properties from experiments with calculated hydraulic properties from image. Soil samples were taken from a site located at the Baltimore science center, which is located inside of the city. Undisturbed columns were taken from the site and scanned with a computer tomographer at resolutions of 22 ${\mu}m$. Pore networks were extracted by medial-axis transformation and were used to measure pore geometry from one of the scanned samples. Water retention and unsaturated hydraulic conductivity values were calculated from the soil image. Properties of soil bulk density, water retention and unsaturated hydraulic conductivity were measured from three replicates of scanned soil samples. 3D image analysis provided accurate detailed pore properties such as individual pore volumes, pore length, and tortuosity of all pores. These data made possible to calculate accurate estimations of water retention and hydraulic conductivity. Combination of the calculated and measured hydraulic properties gave more accurate information on pore sizes over wider range than measured or calculated data alone. We could conclude that the hydraulic property computed from soil images and laboratory measurements can describe a full structure of intra- and inter-aggregate pores in soil.