에이닷의 이미지 이해와 생성 기술 발전 - AI 이미지 대화와 모션 프로필 서비스

Watch a 20-minute technical conference presentation from SK TECH SUMMIT 2023 exploring how A. (A.dot) developed capabilities for image understanding and generation. Learn about the implementation of AI image dialogue features that utilize over 1 billion image-text pairs incorporating Korean regional characteristics, along with techniques to minimize hallucination through proper prompting for natural LLM responses. Discover the AI Motion Profile service that can generate 9 different styles of profile images and animations, including watercolors and magic school themes, while maintaining user facial features. Presented by Yeol-min Baek from SK Telecom's Multimodal team, who shares his journey developing Vision AI technologies from SVM to CNN, Transformer, and Diffusion models, contributing to driving assistance, facial recognition security, and Multimodal AI-based services.