Welcome to Pixel Lab!

Our lab works on a variety of machine learning topics, particularly in Multimodal Learning, Self-supervised Learning, and Social Artificial Intelligence. We take interdisciplinary approaches—spanning computer vision, natural language processing, audio processing, and other relevant fields—to address real-world challenges. For more details on our research areas, please visit here

Lab News

• A paper on Human Gesture Understanding was accepted at NeurIPS 2025
  Toward Human Deictic Gesture Target Estimation 

• A paper on Speech-Driven 3D Facial Animation was accepted at ICCV 2025
  MemoryTalker: Personalized Speech-Driven 3D Facial Animation via Audio-Guided Stylization 

• Our lab has secured a prestigious research grant from NRF (우수중견연구), amounting to 1.2 billion KRW over five years! 

• Five papers on Social Understanding, Multimodal Understanding, and Generative Model were accepted at CVPR 2025
  SocialGesture: Delving into Multi-person Gesture Understanding
  Gaze-LLE: Gaze Target Estimation via Large-Scale Learned Encoders (Highlight, Top 4%)
  Object-aware Sound Source Localization via Audio-Visual Scene Understanding
  Question-Aware Gaussian Experts for Audio-Visual Question Answering (Highlight, Top 4%)
  Unleashing In-context Learning of Autoregressive Models for Few-shot Image Manipulation (Highlight, Top 4%)

• A paper on Video Moment Retrieval & Highlight Detection was accepted at AAAI 2025
  Watch Video, Catch Keyword: Context-aware Keyword Attention for Moment Retrieval and Highlight Detection

• Sangmin was selected as an outstanding reviewer at BMVC 2024

• A paper on Social AI Survey was released on arXiv
  Towards Social AI: A Survey on Understanding Social Interactions

• A paper on Video Point Tracking was accepted at ECCVW 2024
  Leveraging Object Priors for Point Tracking

• A paper on Text-Video Retrieval was accepted in Pattern Recognition
  Text-Guided Distillation Learning to Diversify Video Embeddings for Text-Video Retrieval

• A paper on Social Interaction Modeling was accepted at CVPR 2024 (Oral, Top 0.8%)
  Modeling Multimodal Social Interactions: New Challenges and Baselines with Densely Aligned Representations 

• A paper on Sound Source Localization was accepted at CVPR 2024
  Learning to Visually Localize Sound Sources from Mixtures without Prior Source Knowledge 

• A paper on Speech-Driven 3D Facial Animation was accepted at ICIP 2024
  Analyzing Visible Articulatory Movements in Speech Production for Speech-Driven 3D Facial Animation