A novel framework that redefines motion guidance in video diffusion by extracting 'pseudo-flow' from 2D attention layers, enabling more interpretable and robust motion transfer.
IEIE 2025 Undergraduate Paper Competition (Excellence Award)
We propose CAMEO, a novel framework that aligns correspondence and attention maps to enable consistent multi-view generation from diffusion models.
arXiv preprint
A benchmark dataset and evaluation protocol for object motion transfer, facilitating research in dynamic scene understanding and generation.
ITC-CSCC (Oral)
Developing an automated dataset generation pipeline for GUI motion code with VLMs, bridging the gap between visual design and code implementation.
Finetuned semantic map-conditioned LDMs with ControlNet for unsupervised and unpaired Synthetic-to-Real image translation.
AI profile picture generation service for Korea/Yonsei University students using ID preservation Stable Diffusion, supported by Kakao Brain.
Proposed faster and stable caption generation utilizing Gibbs sampling and Masked VLM for zero-shot image captioning.
Proposed a method for better high-frequency details in NeRF training using adversarial loss and discriminator.
Music recommendation service using VLM and LLM-based lyrics augmentation to match user preferences and context.