Seunghyeon Seo

I'm a Ph.D. student at College of Engineering in Seoul National University, where I'm advised by Prof. Nojun Kwak in the Machine Intelligence and Pattern Analysis Lab (MIPAL). Previously, I got my bachelor's degree from College of Agriculture and Life Sciences in SNU, where I majored in Agricultural Economics.

I'm interested in computer vision, machine learning, and neural rendering. Much of my interest is currently focused on the efficient training framework of NeRF and 3D-GS.

Email  /  CV  /  Google Scholar  /  LinkedIn  /  Github

Apr. 2024: One paper is accepted to CVPR 2024 Workshop on Efficient Large Vision Models.

Feb. 2024: I will be joining Meta Reality Labs as a research scientist intern this summer.

We leverage the pre-trained multimodal model CLIP to achieve state-of-the-art performance in video highlight detection by fine-tuning the encoder and integrating a novel saliency pooling technique.

We cast an hourglass as an additional training ray, which adaptively regularizes the high-frequency components of the samples, and enhance the integrity of training framework by conceptualizing the hourglass as a bundle of flipped diffuse reflection rays, aligning with the Lambertian assumption.

We simplify outdoor scene relighting for NeRF by aligning with the sun, eliminating the need for environment maps and speeding up the process using a novel cubemap concept within the framework of TensoRF.

We expedite ViT inference by concatenating abstract visual tokens from multiple images along dim=1 and processing them collectively.

We utilize the flipped reflection rays as additional training resources for the few-shot novel view synthesis, leading to more accurate surface normal estimation.

We model the high-dimensional joint distribution of human keypoints with a mixture density model by Random Keypoint Grouping strategy and achieve competitive performance working in real-time by eliminating additional instance identification process.

We propose the end-to-end multi-object Detection with a Regularized Mixture Model (D-RMM), which is trained by minimizing the NLL with the proposed regularization term, maximum component maximization (MCM) loss, preventing duplicate predictions.

We model a ray with mixture density model, leading to efficient learning of density distribution with sparse inputs, and propose an effective auxiliary task of ray depth estimation for few-shot novel view synthesis.

We introduce a simple yet effective data augmentation method, Mix/UnMix (MUM), which unmixes feature tiles for the mixed image tiles for the SSOD framework.

Thanks for sharing the website template, Jon Barron. :)