Skip to content
← projects
ai Python · Ollama · LLaVA · Gradio

AI Photo Renamer

A local utility that analyzes images with a vision model through Ollama and assigns descriptive names while grouping similar results.

Problem
Photo folders filled with generic or numeric filenames are difficult to browse and organize without renaming every file by hand.
Solution
The tool sends images to a local vision model, turns each description into a filename, and uses fuzzy similarity and numbering to handle groups and collisions.
Outcome
A private, resumable workflow for producing readable names, with a web interface, dry-run mode, and no cloud API requirement.
Stack
PythonOllamaLLaVAGradio

Local computer vision

Images are encoded and sent to the machine’s Ollama instance. LLaVA is the default model, while other Ollama-compatible vision models can also be selected.

Controlled execution

Dry-run mode makes it possible to review names before applying them. Progress is stored in JSON so interrupted processing can resume without analyzing completed images again.