Hyeonwoo Kim
@Hyeonwoo__Kim
Followers
47
Following
10
Media
6
Statuses
8
MS/PhD student @ SNU VCLab | Human-Object Interaction / Generative Models
Seoul
Joined July 2024
Our code is currently available. Check out our webpage for more details. đź”—Project Page: đź’»Code: @bsw1907, @jhugestar . (4/4).
github.com
Official Repository for ICCV 2025 paper DAViD: Modeling Dynamic Affordance of 3D Objects using Pre-trained Video Diffusion Models - snuvclab/david
0
1
0
DAViD is trained from synthetic 4D HOI samples generated using pre-trained video diffusion models, without laborious capturing. (3/4)
1
0
0
DAViD learns Dynamic Affordance of 3D objects by two key components: (1) human motion model with low-rank adaptation (LoRA) to model human patterns of HOI, and (2) object motion diffusion model conditioned by human motion. (2/4)
1
0
0
Excited to present our work “DAViD” at #ICCV2025!. DAViD is a generative 4D human-object interaction model, which can generate novel HOI motions for various 3D objects, including multi-object interactions. (1/4)
1
6
20
Our code is currently available. Check out our webpage for more details. đź”—Project Page: đź’»Code: @SookwanH, @Patrick920728, @jhugestar. (4/4).
github.com
Official Repository for ECCV 2024 paper Beyond the Contact: Discovering Comprehensive Affordance for 3D Objects from Pre-trained 2D Diffusion Models - snuvclab/coma
0
1
7
ComA is learned from an unbounded set of 3D HOI samples generated using the knowledge of pretrained 2D diffusion models, without laborious annotations. (3/4)
1
0
4
Not only contact patterns, ComA can also model non-contact patterns such as spatial relation and orientational tendency. (2/4)
1
0
1
Thrilled to present our work “ComA” at #ECCV2024 as oral presentation!. ComA (Comprehensive Affordance) is a first approach to model both contact and non-contact affordance for a given 3D object. (1/4)
1
20
100