ArtEvoke integrates several cutting-edge technologies to create a comprehensive platform for connecting art with personal memories:
- Large Multimodal Models (LMMs) for understanding visual and textual content
- Large Language Models (LLMs) for story and description generation, as well as narrative creation
- Vector Database for efficient storage and retrieval of artwork embeddings to enable semantic search
- Embedding Models for semantic understanding of stories and artwork descriptions
- Curated Art Databases including WikiArt, SemArt, and museum collections
The platform operates in two main modes: Memory Reconstruction and Art Exploration. In Memory Reconstruction, users provide a story or memory, and the AI divides it into meaningful sections, then searches for related artworks from curated databases. In Art Exploration, users enter keywords related to themes they want to explore, and the system finds matching artworks, then generates a personalized story connecting the chosen pieces.