About Me
My name is Jaehoon, and I am a senior research engineer at Hyundai Motors in South Korea.
Working on autonomous driving systems has allowed me to solve many real-world problems by leveraging computer vision algorithms.
Challenging myself awards me the most satisfaction in solving real-world problems.
Recently, I have been interested in model compression to efficiently deploy neural network models for hardware platforms with limited computational resources.
Paper
- A Prototype Unit for Image De-raining using Time-Lapse Data
Jaehoon Cho, Jini Yang, Minjung Yoo, Sunok Kim
British Machine Vision Conference (BMVC), Nov. 2024 (accepted)
- Learning Confidence Measure with Transformer in Stereo Matching
Jini Yang, Minjung Yoo, Jaehoon Cho, Sunok Kim
Pattern Recognition (PR), Oct. 2024 (accepted)
- Improving Image de-raining using Reference-guided Transformers
Zihao Ye, Jaehoon Cho, Changjae Oh
IEEE International Conference on Image Processing (ICIP), Oct. 2024
[Project page] [Paper]
- Multi-task Learning for Real-time Autonomous Driving Leveraging Task-adaptive Attention Generator
Wonhyeok Choi, Mingyu Shin, HYUKZAE LEE, Jaehoon Cho, Jaehyeon Park, Sunghoon Im
IEEE International Conference on Robotics and Automation (ICRA), May. 2024
[Paper]
- Memory-Guided Image De-Raining Using Time-Lapse Data
Jaehoon Cho, Seungryong Kim, Kwanghoon Sohn
IEEE Transactions on Image Processing (TIP), vol. 31, pp. 4090 - 4103, Jun. 2022 (Impact factor: 10.856)
[Code] [Paper]
- Wide and Narrow: Video Prediction from Context and Motion
Jaehoon Cho, Jiyoung Lee, Changjae Oh, Wonil Song, Kwanghoon Sohn
British Machine Vision Conference (BMVC), 2021
[Paper]
- Deep Monocular Depth Estimation Leveraging a Large-scale Outdoor Stereo Dataset
Jaehoon Cho, Dongbo Min, Youngjung Kim, Kwanghoon Sohn
Expert Systems With Applications (ESWA), vol. 178, Mar. 2021 (Impact factor: 5.452)
[Dataset] [Paper]
- Pyramid Inter-Attention for High Dynamic Range Imaging
Sungil Choi, Jaehoon Cho, Wonil Song, Jihwan Choe, Jisung Yoo, Kwanghoon Sohn
Sensors, vol. 20, pp. 5102, Jun. 2021 (Impact factor: 3.275)
- Single Image Deraining Using Time-lapse Data
Jaehoon Cho, Seungryong Kim, Dongbo Min, Kwanghoon Sohn
IEEE Transactions on Image Processing (TIP), vol. 29, pp. 7274-7289, Jun. 2020 (Impact factor: 9.340)
[Dataset] [Paper]
- Multi-Task Self-Supervised Visual Representation Learning for Monocular Road Segmentation
Jaehoon Cho, Youngjung Kim, Changjae Oh, Kwanghoon Sohn
IEEE International Conference on Multimedia and Expo (ICME), 2018
[Dataset] [Paper]
Technical Report
- DIML/CVL RGB-D Dataset: 2M RGB-D Images of Natural Indoor and Outdoor Scenes
Jaehoon Cho, Dongbo Min, Youngjung Kim, Kwanghoon Sohn
arXiv, 2021
[Project page] [Paper]
Patents
- 3D object detection using spatial representative knowledge distillation
KR 1020230151997, Jan. 2024.
- Ensemble learning data generation technique for semi-supervised multi-task learning
KR 1020230151996, Jan. 2024.
- Multi-GAN based rain dataset generation technique
KR 1020230178074, Feb. 2024.
- Computationally efficient Argmax, Argmin algorithm
KR 1020230155783, Jan. 2024.
- Multi-task learning method based on the task grouping technique
KR 1020230166997, Jan. 2024.
- GAN-based Night Dataset Generation Techniques Using Histogram Vector
KR 1020230300154, Aug. 2023.
- Semi-supervised semantic segmentation technique using Multiple Teacher Knowledge Distillation
US 18/507,548, DE 102023211571.9, CN 202311628407.4, KR 1020230200237, Sep. 2023.
- Reliable pseudo label generation technique for semi-supervised semantic segmentation
US 18/508,967, KR 1020230044302, Jun. 2023.
- GAN-based dataset generation technique using dual-domain discriminator
KR 1020230100239, Apr. 2023.
- A prototype network-based noise information extraction method
US 18/367,845, KR 1020230034680, May. 2023.
- Multi-task learning techniques using heterogeneous task data
US 18/367,845, KR 1020230035346, May. 2023.
- Method and apparatus for removing noise images with linearity based on deep learning
KR 10-2095444, Mar. 2020.
- DEVICE AND METHOD FOR ESTIMATING ROAD AREA BY DEEP LEARNING USING SELF-SUPERVISED LEARNING
KR 10-2097869, Mar. 2020.
Invited Talks
- CVLAB, “The process of embedded network optimization”, Sep. 2024.
- HMG Developer Conference, “Multi-Task Learning Research Considering Embedded Performance”, 2023. Link
- NAVER LABS, “A Study on Outdoor Scene Understanding in the Dynamic Outdoor Environment”, 2021.
- 42dot, “Deep Neural Network for Single Image De-raining using Real-world Time-lapse Data”, 2021.
Blogs
- HMG DEVELOPERS, “[Deep Learning Optimization] Hardware-Friendly Deep Learning Network: Understanding Reparameterization Techniques”, Feb. 2024 Link
- HMG DEVELOPERS, “[Deep Learning Optimization] The compression/optimization technique of the deep learning model being applied in practice is?”, Apr. 2024 Link
- HMG DEVELOPERS, “[Deep Learning Optimization] Spatial Feature Knowledge Distillation for Monocular 3D Object Detection (1)”, May. 2024 Link)
- HMG DEVELOPERS, “[Deep Learning Optimization] Spatial Feature Knowledge Distillation for Monocular 3D Object Detection (2)”, Jun. 2024 Link
- HMG DEVELOPERS, “Why do we need to conduct quantization?”, Sep. 2024 Link