• The Journal of Korea Robotics Society
• /
• v.18 no.3
• /
• pp.293-298
• /
• 2023

The military is facing a continuous decrease in personnel, and in order to cope with potential accidents and challenges in operations, efforts are being made to reduce the direct involvement of personnel by utilizing the latest technologies. Recently, the use of various sensors related to Manned-Unmanned Teaming and artificial intelligence technologies has gained attention, emphasizing the need for flexible utilization methods. In this paper, we propose four dataset construction methods that can be used for effective training of robots that can be deployed in military operations, utilizing not only RGB image data but also data acquired from IR image sensors. Since there is no publicly available dataset that combines RGB and IR image data, we directly acquired the dataset within buildings. The input values were constructed by combining RGB and IR image sensor data, taking into account the field of view, resolution, and channel values of both sensors. We compared the proposed method with conventional RGB image data classification training using the same learning model. By employing the proposed image data fusion method, we observed improved stability in training loss and approximately 3% higher accuracy.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.10a
• /
• pp.75-77
• /
• 2021

In order to find objects in an image through deep learning technology, an image dataset for learning is required. In order to increase the recognition rate of objects, a large amount of image learning data is required. It is difficult for individuals to build large amounts of datasets because it is expensive. This paper introduces a method for more easily constructing an image dataset including several sides of an object by photographing a rotating image. A method of constructing a dataset by placing an object on a rotating plate, photographing it, and dividing and synthesizing the captured images according to the needs is proposed.

• Journal of the Korea Institute of Military Science and Technology
• /
• v.27 no.2
• /
• pp.167-176
• /
• 2024

Lip-reading is the task of inferring the speaker's utterance from silent video based on learning of lip movements. It is very challenging due to the inherent ambiguities present in the lip movement such as different characters that produce the same lip appearances. Recent advances in deep learning models such as Transformer and Temporal Convolutional Network have led to improve the performance of lip-reading. However, most previous works deal with English lip-reading which has limitations in directly applying to Korean lip-reading, and moreover, there is no a large scale Korean lip-reading dataset. In this paper, we introduce the first large-scale Korean lip-reading dataset with more than 120 k utterances collected from TV broadcasts containing news, documentary and drama. We also present a preprocessing method which uniformly extracts a facial region of interest and propose a transformer-based model based on grapheme unit for sentence-level Korean lip-reading. We demonstrate that our dataset and model are appropriate for Korean lip-reading through statistics of the dataset and experimental results.

• Journal of Drive and Control
• /
• v.19 no.3
• /
• pp.9-15
• /
• 2022

This paper proposes a method for constructing and verifying datasets used in deep learning technology, to prevent safety accidents in automated construction machinery or autonomous vehicles. Although open datasets for developing image recognition technologies are challenging to meet requirements desired by users, this study proposes the interface of virtual simulators to facilitate the creation of training datasets desired by users. The pixel-level training image dataset was verified by creating scenarios, including various road types and objects in a virtual environment. Detecting an object from an image may interfere with the accurate path determination due to occlusion areas covered by another object. Thus, we construct a database, for developing an occlusion area detection algorithm in a virtual environment. Additionally, we present the possibility of its use as a deep learning dataset to calculate a grid map, that enables path search considering occlusion areas. Custom datasets are built using the RDBMS system.

• Wind and Structures
• /
• v.36 no.6
• /
• pp.423-434
• /
• 2023

Aerodynamic force coefficients are generally obtained from traditional wind tunnel tests or computational fluid dynamics (CFD). Unfortunately, the techniques mentioned above can sometimes be cumbersome because of the cost involved, such as the computational cost and the use of heavy equipment, to name only two examples. This study proposed to build a deep neural network model to predict the aerodynamic force coefficients based on data collected from CFD simulations to overcome these drawbacks. Therefore, a series of CFD simulations were conducted using different geometric parameters to obtain the aerodynamic force coefficients, validated with wind tunnel tests. The results obtained from CFD simulations were used to create a dataset to train a multilayer perceptron artificial neural network (ANN) model. The models were obtained using three optimization algorithms: scaled conjugate gradient (SCG), Bayesian regularization (BR), and Levenberg-Marquardt algorithms (LM). Furthermore, the performance of each neural network was verified using two performance metrics, including the mean square error and the R-squared coefficient of determination. Finally, the ANN model proved to be highly accurate in predicting the force coefficients of similar bridge sections, thus circumventing the computational burden associated with CFD simulation and the cost of traditional wind tunnel tests.

• Journal of Broadcast Engineering
• /
• v.27 no.1
• /
• pp.124-132
• /
• 2022

Recently, due to COVID-19, online remote classes and non-face-to-face exams have made it difficult to manage class attitudes and exam cheating. Therefore, there is a need for a system that automatically recognizes and detects the behavior of students online. Action recognition, which recognizes human action, is one of the most studied technologies in computer vision. In order to develop such a technology, data including human arm movement information and information about surrounding objects, which can be key information in online classes and exams, are needed. It is difficult to apply the existing dataset to this system because it is classified into various fields or consists of daily life action. In this paper, we propose a dataset that can classify attitudes in real-time online tests and classes. In addition, it shows whether the proposed dataset is correctly constructed through comparison with the existing action recognition dataset.

• Journal of Construction Engineering and Project Management
• /
• v.4 no.4
• /
• pp.9-16
• /
• 2014

The purpose of this paper is to develop a model for forecasting early design construction cost of building projects using Artificial Neural Network (ANN). Eighty questionnaires distributed among construction organizations were utilized to identify significant parameters for the building project costs. 169 case studies of building projects were collected from the construction industry in Gaza Strip. The case studies were used to develop ANN model. Eleven significant parameters were considered as independent input variables affected on "project cost". The neural network model reasonably succeeded in estimating building projects cost without the need for more detailed drawings. The average percentage error of tested dataset for the adapted model was largely acceptable (less than 6%). Sensitivity analysis showed that the area of typical floor and number of floors are the most influential parameters in building cost.

• International conference on construction engineering and project management
• /
• 2022.06a
• /
• pp.361-369
• /
• 2022

Unsafe actions and behaviors of workers cause most accidents at construction sites. Nowadays, occupational safety is a top priority at construction sites. However, this problem often requires money and effort from investors or construction owners. Therefore, decreasing the accidents rates of workers and saving monitoring costs for contractors is necessary at construction sites. This study proposes an unsafe behavior detection method based on a skeleton model to classify three common unsafe behaviors on the scaffold: climbing, jumping, and running. First, the OpenPose method is used to obtain the workers' key points. Second, all skeleton datasets are aggregated from the temporary size. Third, the key point dataset becomes the input of the action classification model. The method is effective, with an accuracy rate of 89.6% precision and 90.5% recall of unsafe actions correctly detected in the experiment.

• Korean Journal of Construction Engineering and Management
• /
• v.21 no.6
• /
• pp.125-134
• /
• 2020

Natural language encountered in construction documents largely deviates from those that are recommended by the authorities. Such practice that is lacking in coherence will discourage integrated research with automation, and it will hurt the productivity in the industry for the long run. This research aims to compare multiple string similarity (string matching) algorithms to compare each algorithm's performance in recognizing the same task name written in multiple different ways. We also aim to start a debate on how prevalent the aforementioned deviation is. Finally, we composed a small dataset that associates construction task names found in practice with the corresponding task names that are less cluttered w.r.t their formatting. We expect that this dataset can be used to validate future natural language processing approaches.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.60 no.2
• /
• pp.398-406
• /
• 2011

In this paper, we introduce a new architecture of PSO-based Polynomial Neural Networks (PNN) and discuss its comprehensive design methodology. The conventional PNN is based on a extended Group Method of Data Handling (GMDH) method, and utilized the polynomial order (viz. linear, quadratic, and modified quadratic) as well as the number of node inputs fixed (selected in advance by designer) at Polynomial Neurons located in each layer through a growth process of the network. Moreover it does not guarantee that the conventional PNN generated through learning results in the optimal network architecture. The PSO-based PNN results in a structurally optimized structure and comes with a higher level of flexibility that the one encountered in the conventional PNN. The PSO-based design procedure being applied at each layer of PNN leads to the selection of preferred PNs with specific local characteristics (such as the number of input variables, input variables, and the order of the polynomial) available within the PNN. In the sequel, two general optimization mechanisms of the PSO-based PNN are explored: the structural optimization is realized via PSO whereas in case of the parametric optimization we proceed with a standard least square method-based learning. To evaluate the performance of the PSO-based PNN, the model is experimented with using Gas furnace process data, and pH neutralization process data. For the characteristic analysis of the given entire data with non-linearity and the construction of efficient model, the given entire system data is partitioned into two type such as Division I(Training dataset and Testing dataset) and Division II(Training dataset, Validation dataset, and Testing dataset). A comparative analysis shows that the proposed PSO-based PNN is model with higher accuracy as well as more superb predictive capability than other intelligent models presented previously.