Search

Your search keyword '"Kim, Taejoo"' showing total 16 results
Start Over Author "Kim, Taejoo" Publication Year Range Last 3 years
16 results on '"Kim, Taejoo"'

6. INSANet: INtra-INter Spectral Attention Network for Effective Feature Fusion of Multispectral Pedestrian Detection.

Author

Lee, Sangin, Kim, Taejoo, Shin, Jeongmin, Kim, Namil, and Choi, Yukyung

Subjects
*PEDESTRIANS, *THERMOGRAPHY, *DATA augmentation, *COMPUTER vision
Abstract

Pedestrian detection is a critical task for safety-critical systems, but detecting pedestrians is challenging in low-light and adverse weather conditions. Thermal images can be used to improve robustness by providing complementary information to RGB images. Previous studies have shown that multi-modal feature fusion using convolution operation can be effective, but such methods rely solely on local feature correlations, which can degrade the performance capabilities. To address this issue, we propose an attention-based novel fusion network, referred to as INSANet (INtra-INter Spectral Attention Network), that captures global intra- and inter-information. It consists of intra- and inter-spectral attention blocks that allow the model to learn mutual spectral relationships. Additionally, we identified an imbalance in the multispectral dataset caused by several factors and designed an augmentation strategy that mitigates concentrated distributions and enables the model to learn the diverse locations of pedestrians. Extensive experiments demonstrate the effectiveness of the proposed methods, which achieve state-of-the-art performance on the KAIST dataset and LLVIP dataset. Finally, we conduct a regional performance evaluation to demonstrate the effectiveness of our proposed network in various regions. [ABSTRACT FROM AUTHOR]

Published
2024
Full Text
View/download PDF

8. Optimisation of aero-manufacturing characteristics of aircraft ribs

Author

Kim, Taejoo, Kipouros, Timoleon, Brintrup, Alexandra, Farnfield, James, Di Pasquale, Davide, Kipouros, Timoleon [0000-0003-3392-283X], Brintrup, Alexandra [0000-0002-4189-2434], and Apollo - University of Cambridge Repository

Subjects
Machine Learning, Manufacturing, Aerospace Engineering, Optimisation
Abstract

The main purpose of this study was to combine the currently separate objectives of aerodynamic performance and manufacturing efficiency, then find an optimal point of operation for both objectives. An additional goal of the study was to explore the effects of changes in design features, the position of the spars, and analyse how the changes influenced the optimal operating conditions. A machine-learning approach was taken to combine and model the gathered aero-manufacturing data, and a multi-objective optimisation approach utilising genetic algorithms was implemented to find the trade-off relationship between optimal target objectives (mission performance and manufacturability). The main achievements and findings of the study were: The study was a success in building a machine-learning model for the combined aero-manufacturing data utilising software library XGBoost; multi-objective optimisation, which did not include spar positions as a variable found the trade-off region between high manufacturability and high mission performance, with choices that offered reasonably high values of both; there was no clearly identified correlation between a small change in spar position and the target objectives; multi-objective optimisation with spar positions resulted in a trade-off relationship between target objectives, which was different from the trade-off relationship found in optimisation without spar positions; multi-objective optimisation with spar positions also offered more flexibility in the choice of manufacturing processes available for a given design; and the range of bump amplitudes for solutions found by multi-objective-optimisation with spar positions was lower and more focused than those found by optimisation without spar positions.

Published
2022

15. Surrogate modeling of the fan plot of a rotor system considering composite blades using convolutional neural networks with image composition

Author

Noh, Hong-Kyun, Lim, Jae Hyuk, Lee, Seungchul, Kim, Taejoo, and Kim, Deog-Kwan

Abstract

This study proposes an image composition technique based on convolutional neural networks (CNNs) to construct a surrogate model for predicting fan plots of three-dimensional (3D) composite blades, which represent natural frequency lists at different rotational speeds. The proposed method composes critical 2D cross-section images to improve the accuracy of the model. Numerical examples with various compositions of cross-section images are presented to demonstrate the efficacy of the CNN model. Additionally, gradient-weighted class activation mapping analysis is used to reveal the relationship between the internal structure of the blade and the fan plots. The study shows that using multiple images in the image composition technique improves the accuracy of the model compared to using single or fewer images. Overall, the proposed method provides a promising approach for predicting fan plots of 3D composite blades using CNN models.Graphical Abstract

Published
2023
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results