Bumsoo Kim
Korea University
9 Papers
41 Citations
Bumsoo Kim is an academic researcher from Korea University. The author has contributed to research in topics: Object detection & Deep learning. The author has an hindex of 6, co-authored 9 publications.
Chat about Author
Papers
HOTR: End-to-End Human-Object Interaction Detection with Transformers
Bumsoo Kim,Junhyun Lee,Jaewoo Kang,Eun-Sol Kim,Hyunwoo Kim +4 more
- 20 Jun 2021
TL;DR: Zhang et al. as mentioned in this paper presented a novel framework, referred by HOTR, which directly predicts a set of human, object, interaction triplets from an image based on a transformer encoder-decoder architecture.
UnionDet: Union-Level Detector Towards Real-Time Human-Object Interaction Detection
Bumsoo Kim,Taeho Choi,Jaewoo Kang,Hyunwoo Kim +3 more
- 23 Aug 2020
TL;DR: UnionDet, a one-stage meta-architecture for HOI detection powered by a novel union-level detector that eliminates this additional inference stage by directly capturing the region of interaction and shows a significant reduction in interaction prediction time.
Detection of masses in mammograms using a one-stage object detector based on a deep convolutional neural network.
Hwejin Jung,Bumsoo Kim,Inyeop Lee,Minhwan Yoo,Junhyun Lee,Soo Youn Ham,Okhee Woo,Jaewoo Kang +7 more
TL;DR: This work proposes a mass detection model based on RetinaNet that achieves comparable or better performance than more complex state-of-the-art models including the two-stage object detector and makes the model’s weights pre-trained on both GURO and INbreast publicly available.
Classification of lung nodules in CT scans using three-dimensional deep convolutional neural networks with a checkpoint ensemble method
TL;DR: This paper introduces a high performance nodule classification method that uses three dimensional deep convolutional neural networks (DCNNs) and an ensemble method to distinguish nodules between non-nodules and achieves higher competition performance metric (CPM) scores than the state-of-the-art methods using deep learning.
Hemodynamic energy generated by a combined centrifugal pump with an intra-aortic balloon pump.
Choon Hak Lim,Ho Sung Son,Yung Hu Fang,Jung Joo Lee,Kwang Je Baik,Kyung Hyun Kim,Bumsoo Kim,Hye Won Lee,Kyung Sun +8 more
TL;DR: In a cardiac-arrested animal model, pulsatility generated by a IABP/CP combination may be effective in terms of energy equivalent pressure and surplus hemodynamic energy.
23