• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2022년도 추계학술발표대회
• /
• pp.478-480
• /
• 2022

This paper studies human activity image classification using deep transfer learning techniques focused on the inception convolutional neural networks (InceptionV3) model. For this, we used UFC-101 public datasets containing a group of students' behaviors in mathematics classrooms at a school in Thailand. The video dataset contains Play Sitar, Tai Chi, Walking with Dog, and Student Study (our dataset) classes. The experiment was conducted in three phases. First, it extracts an image frame from the video, and a tag is labeled on the frame. Second, it loads the dataset into the inception V3 with transfer learning for image classification of four classes. Lastly, we evaluate the model's accuracy using precision, recall, F1-Score, and confusion matrix. The outcomes of the classifications for the public and our dataset are 1) Play Sitar (precision = 1.0, recall = 1.0, F1 = 1.0), 2), Tai Chi (precision = 1.0, recall = 1.0, F1 = 1.0), 3) Walking with Dog (precision = 1.0, recall = 1.0, F1 = 1.0), and 4) Student Study (precision = 1.0, recall = 1.0, F1 = 1.0), respectively. The results show that the overall accuracy of the classification rate is 100% which states the model is more powerful for learning UCF-101 and our dataset with higher accuracy.

• Geomechanics and Engineering
• /
• 제36권3호
• /
• pp.231-245
• /
• 2024

In recent decades, the protection and vulnerability of civil structures under explosion loads became a critical issue in terms of security, which may cause loss of lives and structural damage. Concrete retaining walls also restrict soils and slopes from displacements; meanwhile, intensive temporary loading may cause massive damage. In the current study, the modified Johnson-Holmquist (also known as J-H2) material model is implemented for concrete materials to model damages into the ABAQUS through user-subroutines to predict the blasting-induced concrete damages and volume strains. For this purpose, a 3D finite-element model of the concrete retaining wall was conducted in coupled Eulerian-Lagrangian simulation. Subsequently, a blast load equal to 500 kg of TNT was considered in three different positions due to UFC 3-340-02. Influences of the critical parameters in smooth blastings, such as distance from a free face, position, and effective blasting time, on concrete damage rate and destroy patterns, are explored. According to the simulation results, the concrete penetration pattern at the same distance is significantly influenced by the density of the progress environment. The result reveals that the progress of waves and the intensity of damages in free-air blasting is entirely different from those that progress in a dense surrounding atmosphere such as soil. Half-damaged elements in air blasts are more than those of embedded explosions, but dense environments such as soil impose much more pressure in a limited zone and cause more destruction in retaining walls.

• 한국건축시공학회지
• /
• 제14권6호
• /
• pp.615-621
• /
• 2014

본 연구는 수치해석과 폭발 실증시험을 통해 조립식 PC 아치 탄약고의 설계 및 시공도면을 제시하였다. 제안된 탄약고의 방호성능은 미군의 UFC 3-340-02와 한국군의 국방 군사시설 설계기준(탄약고 설계지침)에서 제시된 최대 회전각도로 판별하였다. 아치형 탄약고의 기초설계를 위해 곡면형태인 반원형과 오발형에 대하여 폭발하중에 대한 방호성능을 수치해석을 통해 살펴본 결과 오발형이 타원형에 비하여 지점회전각이 적어 전반적으로 방호성능이 뛰어남을 확일 할 수 있었다. 이를 통해 아치형 조립식 PC 탄약고의 설계안 및 시공도를 제안하였으며, TNT 133.75kg을 구조체 벽체로부터 00m 이격하여 폭발시켜 실제 시공된 탄약고에 대한 방호성능을 검증하였다. 결론적으로, 본 연구에서 제안된 탄약고는 충분한 방호 여유력을 확보하고 있는 것으로 확인할 수 있었다. 향후 이러한 방호 여유력을 고려하여 재료의 절감을 통한 보다 경제적인 탄약고 설계도 가능할 것으로 사료된다.

• 한국안전학회지
• /
• 제31권3호
• /
• pp.8-15
• /
• 2016

The purpose of this study is to design a model with the structural stability so as not to lose the operational function due to structural plastic or fail of a sliding blast door by blast pressure to this aim, a numerical simulation was performed using full-size experiments and M&S (Modeling & Simulation) of the sliding blast door. The sliding blast door ($W3,000{\times}H2,500mm$) under the blast load is in the form of a sliding type 2-way metal grill, which was applied by a design blast pressure (reflected pressure $P_r$) of 17 bar. According to the experimental results of a real sliding blast door under blast load, the blast pressure reached the sliding blast door approximately 4.3 ms after the explosion and lasted about 4.0 ms thereafter. The maximum blast pressure($P_r$) was 347.7 psi (2,397.3 kPa), it is similar to the UFC 3-340-02 of Parameter(91 %). In addition, operation inspection that was conducted for the sliding blast door after real test showed a problem of losing the door opening function, which was because of the fail of the Reversal Bolt that was installed to prevent the shock due to rebound of the blast door from the blast pressure. According to the reproduction of the experiment through M&S by applying the blast pressure measurement value of the full-size experiments, the sliding blast door showed a similar result to the full-size experiment in that the reversal bolt part failed to lose the function. In addition, as the pressure is concentrated on the failed reversal bolt, the Principal Tensile Failure Stress was exceeded in only 1.25 ms after the explosion, and the reversal bolt completely failed after 5.4 ms. Based on the result of the failed reversal bolt through the full-size experiment and M&S, the shape and size of the bolts were changed to re-design the M&S and re-analyze the sliding blast door. According to the M&S re-analysis result when the reversal bolt was designed in a square of 25 mm ($625mm^2$), the maximum pressure that the reversal bolt receives showed 81% of the principal tensile failure stress of the material, in plastic stage before fail.