• Journal of Korean Tunnelling and Underground Space Association
• /
• v.24 no.6
• /
• pp.513-524
• /
• 2022

Recently, detecting damages of civil infrastructures from digital images using deep learning technology became a very popular research topic. In order to adapt those methodologies to the field, it is essential to explain robustness of deep learning models. Our research points out that the existing pixel-based deep learning model evaluation metrics are not sufficient for detecting cracks since cracks have linear appearance, and proposes a new evaluation methodology to explain crack segmentation deep learning model more rationally. Specifically, we design, implement and validate a methodology to generate tolerance buffer alongside skeletonized ground truth data and prediction results to consider overall similarity of topology of the ground truth and the prediction rather than pixel-wise accuracy. We could overcome over-estimation or under-estimation problem of crack segmentation model evaluation through using our methodology, and we expect that our methodology can explain crack segmentation deep learning models better.

• Korean Journal of Remote Sensing
• /
• v.37 no.5_1
• /
• pp.847-859
• /
• 2021

Recently, as the number of high-resolution satellite SAR images increases, the demand for precise matching of SAR imagesin change detection and image fusion is consistently increasing. RMSE (Root Mean Square Error) values using GCPs (Ground Control Points) selected by analysts have been widely used for quantitative evaluation of image registration results, while it is difficult to find an approach for automatically measuring the registration accuracy. In this study, a feasibility analysis was conducted on using the FSIM (Feature Similarity) index as a measure to evaluate the registration accuracy. TerraSAR-X (TSX) staring spotlight data collected from various incidence angles and orbit directions were used for the analysis. FSIM was almost independent on the spatial resolution of the SAR image. Using a single SAR image, the FSIM with respect to registration errors was analyzed, then use it to compare with the value estimated from TSX data with different imaging geometry. FSIM index slightly decreased due to the differencesin imaging geometry such as different look angles, different orbit tracks. As the result of analyzing the FSIM value by land cover type, the change in the FSIM index according to the co-registration error was most evident in the urban area. Therefore, the FSIM index calculated in the urban was mostsuitable for determining the accuracy of image registration. It islikely that the FSIM index has sufficient potential to be used as an index for the co-registration accuracy of SAR image.

• Journal of the Korean earth science society
• /
• v.39 no.6
• /
• pp.533-544
• /
• 2018

Rainfall is one of the most important meteorological variables in meteorology, agriculture, hydrology, natural disaster, construction, and architecture. Recently, satellite remote sensing is essential to the accurate detection, estimation, and prediction of rainfall. In this study, the accuracy of Integrated Multi-satellite Retrievals for GPM (IMERG) product, a composite rainfall information based on Global Precipitation Measurement (GPM) satellite was evaluated with ground observation data in the inland of Korea. The Automatic Weather Station (AWS)-based rainfall measurement data were used for validation. The IMERG and AWS rainfall data were collocated and compared during one year from January 1, 2016 to December 31, 2016. The coastal regions and islands were also evaluated irrespective of the well-known uncertainty of satellite-based rainfall data. Consequently, the IMERG data showed a high correlation (0.95) and low error statistics of Bias (15.08 mm/mon) and RMSE (30.32 mm/mon) in comparison to AWS observations. In coastal regions and islands, the IMERG data have a high correlation more than 0.7 as well as inland regions, and the reliability of IMERG data was verified as rainfall data.

• Korean Journal of Remote Sensing
• /
• v.34 no.6_2
• /
• pp.1273-1282
• /
• 2018

Recently, active research on the Arctic Ocean has been conducted due to the influence of global warming and new Arctic ship route. Although previous studies already calculated quantitative extent of sea ice using passive microwave radiometers, melting at the edge of sea ice and surface roughness were hardly considered due to low spatial resolution. Since Sentienl-1A/B data in Extended Wide (EW) mode are being distributed as free of charge and bulk data for Arctic sea can be generated during a short period, the entire Arctic sea ice data can be covered in high spatial resolution by mosaicking bulk data. However, Sentinel-1A/B data in EW mode, especially in HV polarization, needs significant radiometric correction for further classification. Thus, in this study, we developed algorithms that can correct thermal noise and scalloping effects, and confirmed that Arctic sea ice and open-water were well classified using the corrected dual-polarization SAR data.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.20 no.6
• /
• pp.85-91
• /
• 2020

Recently, video-taping equipment such as CCTV have been seeing more use for crime prevention and general safety concerns. Since these video-taping equipment operates all throughout the day, the need for security personnel is lessened, and naturally costs incurred from managing such manpower should also decrease. However, technology currently used predominantly lacks self-sufficiency when given the task of searching for a specific object in the recorded video such as a person, and has to be done manually; current security-based video equipment is insufficient in an environment where real-time information retrieval is required. In this paper, we propose a technology that uses the latest deep-learning technology and OpenCV library to quickly search for a specific person in a video; the search is based on the clothing information that is inputted by the user and transmits the result in real time. We implemented our system to automatically recognize specific human objects in real time by using the YOLO library, whilst deep learning technology is used to classify human clothes into top/bottom clothes. Colors are also detected through the OpenCV library which are then all combined to identify the requested object. The system presented in this paper not only accurately and quickly recognizes a person object with a specific clothing, but also has a potential extensibility that can be used for other types of object recognition in a video surveillance system for various purposes.

• Geophysics and Geophysical Exploration
• /
• v.24 no.3
• /
• pp.73-77
• /
• 2021

Induced polarization (IP) is useful for mineral exploration and hydrogeological studies by visualizing the electrochemical reactions at the interface between polarized minerals and groundwater. Frequency-domain IP (FDIP) is not actively applied to field surveys because it takes longer to acquire data, despite its higher data quality than conventional time-domain IP. However, data quality is more important in current mineral exploration as the targets gradually shift to deep or low-grade ore bodies. In addition, the measurement time reduced by automated instrumentation increases the potential for FDIP field applications. Therefore, we demonstrate that FDIP can detect mineral exploration targets by performing geophysical logging in the boreholes of a skarn deposit, in South Korea. Alternating current (AC) resistivity, percent frequency effect (PFE) and metal factor (MF) were calculated from impedance values obtained at two different frequencies. Skarn zones containing magnetite or pyrite showed relatively low AC resistivity, high PFE, and high MF compared to other zones. Therefore, FDIP surveys are considered to be useful for mineral exploration.

• Korean Journal of Agricultural and Forest Meteorology
• /
• v.24 no.2
• /
• pp.95-114
• /
• 2022

This study presented the possibility of automatic frost observation and the related image data acquisition through the design and installation of a Multiple-sensor based Frost Observation System (MFOS). The MFOS is composed of an RGB camera, a thermal camera and a leaf wetness sensor, and each device performs complementary roles. Through the test operation of the equipment before the occurrence of frost, the voltage value of the leaf wetness sensor increased when maintaining high relative humidity in the case of no precipitation. In the case of Gapyeong- gun, the high relative humidity was maintained due to the surrounding agricultural waterways, so the voltage value increased significantly. In the RGB camera image, leaf wetness sensor and the surface were not observed before sunrise and after sunset, but were observed for the rest of the time. In the case of precipitation, the voltage value of the leaf wetness sensor rapidly increased during the precipitation period and decreased after the precipitation was terminated. In the RGB camera image, the leaf wetness sensor and surface were observed regardless of the precipitation phenomenon, but the thermal camera image was taken due to the precipitation phenomenon, but the leaf wetness sensor and surface were not observed. Through, where actual frost occurred, it was confirmed that the voltage value of leaf wetness sensor was higher than the range corresponding to frost, but frost was observed on the surface and equipment surface by the RGB camera.

• Journal of the Institute of Convergence Signal Processing
• /
• v.23 no.3
• /
• pp.166-172
• /
• 2022

In this study, we photograph driving obstacle objects such as crosswalks, side spheres, manholes, braille blocks, partial ramps, temporary safety barriers, stairs, and inclined curb that hinder or cause inconvenience to the movement of the vulnerable using electric mobility aids. We develop an optimal AI model that classifies photographed objects and automatically recognizes them, and implement an algorithm that can efficiently determine obstacles in front of electric mobility aids. In order to enable object detection to be AI learning with high probability, the labeling form is labeled as a polygon form when building a dataset. It was developed using a Mask R-CNN model in Detectron2 framework that can detect objects labeled in the form of polygons. Image acquisition was conducted by dividing it into two groups: the general public and the transportation weak, and image information obtained in two areas of the test bed was secured. As for the parameter setting of the Mask R-CNN learning result, it was confirmed that the model learned with IMAGES_PER_BATCH: 2, BASE_LEARNING_RATE 0.001, MAX_ITERATION: 10,000 showed the highest performance at 68.532, so that the user can quickly and accurately recognize driving risks and obstacles.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.23 no.3
• /
• pp.199-206
• /
• 2023

This study utilizes deep learning algorithms to automatically classify construction waste into three categories: wood waste, plastic waste, and concrete waste. Two models, VGG-16 and ViT (Vision Transformer), which are convolutional neural network image classification algorithms and NLP-based models that sequence images, respectively, were compared for their performance in classifying construction waste. Image data for construction waste was collected by crawling images from search engines worldwide, and 3,000 images, with 1,000 images for each category, were obtained by excluding images that were difficult to distinguish with the naked eye or that were duplicated and would interfere with the experiment. In addition, to improve the accuracy of the models, data augmentation was performed during training with a total of 30,000 images. Despite the unstructured nature of the collected image data, the experimental results showed that VGG-16 achieved an accuracy of 91.5%, and ViT achieved an accuracy of 92.7%. This seems to suggest the possibility of practical application in actual construction waste data management work. If object detection techniques or semantic segmentation techniques are utilized based on this study, more precise classification will be possible even within a single image, resulting in more accurate waste classification

• Journal of Aerospace System Engineering
• /
• v.16 no.1
• /
• pp.49-55
• /
• 2022

The role of drones has been expanded to various fields such as agriculture, construction, and logistics. In particular, agriculture drones are emerging as an effective alternative to solve the problem of labor shortage and reduce the input cost. In this study therefore, we proposed the fruit recognition algorithm and harvesting mechanism for fruit harvesting drone system that can safely harvest fruits at high positions. In the fruit recognition algorithm, we employ "You-Only-Look-Once" which is a deep learning-based object detection algorithm and verify its feasibility by establishing a virtual simulation environment. In addition, we propose the fruit harvesting mechanism which can be operated by a single driving motor. The rotational motion of the motor is converted into a linear motion by the scotch yoke, and the opened gripper moves forward, grips a fruit and rotates it for harvesting. The feasibility of the proposed mechanism is verified by performing Multi-body dynamics analysis.