🛠️ NPU SDK Engineer

• Engineered and deployed a Python and C++ based functional simulator for an image NPU, incorporating customized floating-point operations and CUDA-based parallelization to optimize performance.
• Developed a performance simulator for an image NPU that predicts the range of cycles required depending on the network and image size.
• Contributed to research on fisheye image segmentation using general datasets, including literature review, experimentation, and manuscript preparation.

👁 Computer Vision Researcher

• Investigated computer vision, deep learning, scene understanding, and efficient neural networks.

Contact & Channels

👾 Github: https://github.com/seongwn

📚 Google Scholar: https://scholar.google.com/citations?user=ll0Gzb4AAAAJ&hl=ko

Research Interests

Computer Vision, Deep Learning, Scene Understanding, Open World Vision, Efficient Neural Networks

Research Experiences

📚 Publications 논문

• Seongwon Jeong*, Jiyeong Kim*, Sungheui Kim, Dongbo Min. Revisiting Domain-Adaptive Semantic

  Segmentation via Knowledge Distillation. IEEE Transactions on Image Processing (TIP), 2024.

• Sunho Ki, Jinhong Park, Seongwon Jeong, Sureum Choi, Jinwoo Kim, Jeongho Woo. General Dataset Based Image Segmentation for the Fisheye Surveillance Camera Application. International Confer- ence on Computer Graphics and Image Processing (CGIP), 2024.

• Hyesong Lee, Hunsang Lee, Seongwon Jeong, Dongbo Min. Environment Agnostic RepreResentation for Visual Reinforcement Learning. IEEE Int. Conf. on Computer Vision (ICCV), 2023.

• Yujeong Lee*, Seongwon Jeong*, Dongbo Min. Extending Confidence Thresholding to Unsupervised Domain Adaptation for Semantic Segmentation. 2022 Workshop for Image Processing and Image Un- derstanding (IPIU), 2022.

[1] *: equal contribution

[2] Please visit [here] for more information about the publications.

📘 Main Research Experiences 주요 연구 경험