Tae-Jung Kim

Tae-Jung Kim

Auto-generated profile, based on publicly available information

Tae-Jung Kim

Department of Intelligent Systems and Robotics, Chungbuk National University, Cheongju-si, Chungcheongbuk-do, Republic of Korea

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
27 August 2025 in IEEE Access
Verified
Journal Article
Open access
TKTae-Jung Kim
Tae-Jung Kim
TPTae-Hyoung Park
Tae-Hyoung Park
In this paper, a Diffusion-based Involution AttentionYOLO network and a Diffusion-based Four-channel Fusion module are proposed, both specifically for oriented bounding box in overhead imagery. The propoased fusion module uses the Marigold depth-estimation network to infer depth information from a single RGB image, producing a four-channel RGBD input. In the proposed network backbone, involution captures fine-grained spatial relationships across color and depth patterns, and the Convolutional Block Attention Module injects the most salient features and locations into the detection head. Experimental results demonstrate significant improvements across multiple datasets. On PCB data, DIA-YOLO achieves 91.6% improvement in mAP50-95 (32.0% to 61.3%) with 16.5% faster inference time (0.71s). Cross-dataset evaluation shows consistent improvements: 21.7% in PCBself, 11.8% in PCBopen, 31.3% in DOTA, and 7.6% in AITOD. Comparative analysis confirms superiority with 79.32% mAP50 onDOTAand 89.9% mAP50 on AITOD. As the inference time is 0.71s—well below 1 second—this method satisfies real-time requirements for practical overhead imagery applications, providing an effective solution for PCB inspection, satellite analysis, and autonomous systems.

25 February 2025 in IEEE Access
Verified
Journal Article
Open access
TKTae-Jung Kim
Tae-Jung Kim
MHMin-Ho Ha
Min-Ho Ha
SASaba Arshad
Saba Arshad
TPTae-Hyoung Park
Tae-Hyoung Park
In the AI-driven computer vision industry, height measurement of Printed Circuit Board images typically relies on laser or Moiré methods. In this paper, we focus on the Moiré method, known for its high accuracy and fast measurement speed. However, when using Moiré method, shadows and light reflections are generated on Printed Circuit Board surface that cause significant errors in height measurement. To address this problem, we propose a Moiré Inpainting Network, which integrates the Moiré method with an image inpainting model architecture. Our approach leverages a Generative Adversarial Network to accurately identify and reconstruct shadow and reflection regions. The network takes 2D Printed Circuit Board Moiré images as input and outputs heights of Printed Circuit Board. We evaluate performance using Height Reconstruction Rate, Shadow Reconstruction Rate, and Reflection Reconstruction Rate, metrics we define in this paper. Comparative experiments show that our method outperforms state-of-the-art inpainting models for Moiré images, proving its effectiveness in computer vision applications. Moreover, we achieve a reasonable inference time, enabling real-time deployment in Printed Circuit Board manufacturing.

4 November 2024 in IEEE Photonics Journal
Verified
Journal Article
Open access
MHMin-Ho Ha
Min-Ho Ha
SASaba Arshad
Saba Arshad
TKTae-Jung Kim
Tae-Jung Kim
TPTae-Hyoung Park
Tae-Hyoung Park
This paper focuses on three-dimensional measurement system of printed circuit board components. Most automated optical inspections suffer from shadows and reflections obscured by the components. We propose a new method to restore shadow and reflection regions of components for accurate shape measurements, which defines shadow and reflection regions for restoration. The region is restored with a variational auto encoder-based restoration network. It is efficient to input information about both shadow and reflection region to restore the region. Additionally, long skip connection is added to VAE for improved accuracy. We confirmed its superiority through experiments by comparing its performance with other restoration networks such as cycle generative adversarial networks and coherent semantic attention inpainting.