Seungryong Kim

A new method enhances 3D model generation from text prompts.

A tool that personalizes images from text while preserving reference details.

Addressing challenges in federated learning due to diverse devices and data.

RAIN-GS improves 3DGS performance with better initialization and filtering techniques.

An overview of strategies for improving dense matching accuracy.

PAG improves image quality without extra training or prompts, enhancing generation techniques.

MoDiTalker offers improved quality and speed in creating realistic talking head videos.

Talk3D technology enhances lifelike talking head video creation with audio synchronization.

GaussianTalker transforms digital interaction with lifelike talking heads.

GenWarp generates new views from single images while preserving essential details.

Diff-ID enhances person recognition by generating diverse training images.

A new method enhances point tracking accuracy and efficiency in video processing.

A new method improves object recognition using masks without detailed labels.

A new dataset that improves video object tracking accuracy.

Discover how AM-Adapter changes images while keeping key details intact.

New techniques boost image quality from noise without guidance.

Explore how machines analyze images from different angles for better interpretation.