🔑 Documentation Menu▼

Local AI & Semantic Search

Zemoria leverages advanced on-device AI to understand the semantic meaning of your photos rather than relying on basic file tags.

You can search your photo library using natural, descriptive sentences:

  • "a red sports car under the sunset"
  • "dog playing on a snowy field"
  • "happy family eating dinner"

Local Face Recognition

Zemoria runs a lightweight facial embedding model locally. It groups similar faces together in the “People” tab.

  • No cloud uploads: All face encodings are saved in your local SQLite/DBreeze database.
  • Rename and merge: You can name people and merge duplicate clusters.

Under the Hood

We utilize a combination of a light multi-modal representation model (SigLIP) and a compact language model to interpret search queries.

  • DirectML acceleration: On Windows, Zemoria uses DirectML to leverage your GPU (NVIDIA, AMD, or Intel) for sub-second search times.