Jinhyun Kim
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Jinhyun Kim has a total of 40 co-authors. Here are the top co-authors by number of shared publications:
- Sunghoon Kwon
Department of Electrical and Computer Engineering, Seoul National University, Seoul 08826, Republic of Korea
2 shared publications Sungsik Kim
Interdisciplinary Program in Bioengineering, Seoul National University, Seoul, Republic of Korea
2 shared publicationsAhyoun Choi
Interdisciplinary Program in Bioengineering, Seoul National University, Seoul, Republic of Korea
1 shared publication- Amos Chungwon Lee
Bio-MAX Institute, Seoul National University, Seoul, Republic of Korea
1 shared publication - Dohun Kim
Division of Data Science, College of Intelligent Software Convergence, The University of Suwon, Hwaseong 18323, Republic of Korea
1 shared publication - Dong Soon Lee
Department of Laboratory Medicine, Seoul National University College of Medicine, Seoul, Korea
1 shared publication
Latest Publications
- Open Access
- 27 January 2025
- Published by MDPI AG in Electronics
- Vol. 14 (3) , 517
FeatherFace: Robust and Lightweight Face Detection via Optimal Feature Integration
Face detection in resource-constrained environments presents challenges, due to the computational demands of state-of-the-art models and the complexity of real-world conditions, such as variations in scale, pose, and occlusion. This study introduces FeatherFace, a lightweight face-detection architecture with only 0.49 M parameters, designed for high accuracy and efficiency in such environments. Leveraging MobileNet-0.25 as a backbone, FeatherFace incorporates advanced feature-integration strategies, including a bidirectional feature pyramid network (BiFPN), a convolutional block attention module (CBAM), deformable convolutions, and channel shuffling. Evaluated on the WIDERFace dataset, FeatherFace achieves an overall average precision (AP) of 87.2%, with notable performance gains of 4.0% AP on the Hard subset compared with the baseline. Ablation studies highlight the critical role of multiscale feature integration and the strategic placement of attention mechanisms in addressing detection challenges such as small or occluded faces. With its compact design and reduced inference time, FeatherFace bridges the gap between the reliability of computationally intensive models and the need for deploying robust models in highly resource-constrained environments, such as edge devices and embedded systems. This work provides valuable insights for developing robust and lightweight models suited to challenging real-world applications.
- 27 January 2025
- Published by MDPI AG in Electronics
- Vol. 14 (3) , 517
Open AccessFeatherFace: Robust and Lightweight Face Detection via Optimal Feature Integration
Face detection in resource-constrained environments presents challenges, due to the computational demands of state-of-the-art models and the complexity of real-world conditions, such as variations in scale, pose, and occlusion. This study introduces FeatherFace, a lightweight face-detection architecture with only 0.49 M parameters, designed for high accuracy and efficiency in such environments. Leveraging MobileNet-0.25 as a backbone, FeatherFace incorporates advanced feature-integration strategies, including a bidirectional feature pyramid network (BiFPN), a convolutional block attention module (CBAM), deformable convolutions, and channel shuffling. Evaluated on the WIDERFace dataset, FeatherFace achieves an overall average precision (AP) of 87.2%, with notable performance gains of 4.0% AP on the Hard subset compared with the baseline. Ablation studies highlight the critical role of multiscale feature integration and the strategic placement of attention mechanisms in addressing detection challenges such as small or occluded faces. With its compact design and reduced inference time, FeatherFace bridges the gap between the reliability of computationally intensive models and the need for deploying robust models in highly resource-constrained environments, such as edge devices and embedded systems. This work provides valuable insights for developing robust and lightweight models suited to challenging real-world applications.
- Open Access
- 29 August 2023
- Published by Springer Nature in Nature Communications
- Vol. 14 (1) , 1-18
Barcoded multiple displacement amplification for high coverage sequencing in spatial genomics
Determining mutational landscapes in a spatial context is essential for understanding genetically heterogeneous cell microniches. Current approaches, such as Multiple Displacement Amplification (MDA), offer high genome coverage but limited multiplexing, which hinders large-scale spatial genomic studies. Here, we introduce barcoded MDA (bMDA), a technique that achieves high-coverage genomic analysis of low-input DNA while enhancing the multiplexing capabilities. By incorporating cell barcodes during MDA, bMDA streamlines library preparation in one pot, thereby overcoming a key bottleneck in spatial genomics. We apply bMDA to the integrative spatial analysis of triple-negative breast cancer tissues by examining copy number alterations, single nucleotide variations, structural variations, and kataegis signatures for each spatial microniche. This enables the assessment of subclonal evolutionary relationships within a spatial context. Therefore, bMDA has emerged as a scalable technology with the potential to advance the field of spatial genomics significantly.
- 29 August 2023
- Published by Springer Nature in Nature Communications
- Vol. 14 (1) , 1-18
Open AccessBarcoded multiple displacement amplification for high coverage sequencing in spatial genomics
Determining mutational landscapes in a spatial context is essential for understanding genetically heterogeneous cell microniches. Current approaches, such as Multiple Displacement Amplification (MDA), offer high genome coverage but limited multiplexing, which hinders large-scale spatial genomic studies. Here, we introduce barcoded MDA (bMDA), a technique that achieves high-coverage genomic analysis of low-input DNA while enhancing the multiplexing capabilities. By incorporating cell barcodes during MDA, bMDA streamlines library preparation in one pot, thereby overcoming a key bottleneck in spatial genomics. We apply bMDA to the integrative spatial analysis of triple-negative breast cancer tissues by examining copy number alterations, single nucleotide variations, structural variations, and kataegis signatures for each spatial microniche. This enables the assessment of subclonal evolutionary relationships within a spatial context. Therefore, bMDA has emerged as a scalable technology with the potential to advance the field of spatial genomics significantly.
- 1 June 2021
- Published by Elsevier in Biochemical and Biophysical Research Communications
- Vol. 558, 209-215
Phenotype-based single cell sequencing identifies diverse genetic subclones in CD133 positive cancer stem cells
- 1 June 2021
- Published by Elsevier in Biochemical and Biophysical Research Communications
- Vol. 558, 209-215
Phenotype-based single cell sequencing identifies diverse genetic subclones in CD133 positive cancer stem cells
- Open Access
- 1 August 2019
- Published by Elsevier in Clinical Lymphoma Myeloma and Leukemia
- Vol. 19 (8) , e496-e505
Characteristics of Waldenström Macroglobulinemia in Korean Patients According to Mutational Status of MYD88 and CXCR4: Analysis Using Ultra-Deep Sequencing
- 1 August 2019
- Published by Elsevier in Clinical Lymphoma Myeloma and Leukemia
- Vol. 19 (8) , e496-e505
Open AccessCharacteristics of Waldenström Macroglobulinemia in Korean Patients According to Mutational Status of MYD88 and CXCR4: Analysis Using Ultra-Deep Sequencing