FoodMapper
Match free-text food descriptions to standardized reference databases using on-device ML models powered by MLX on Apple Silicon.
Built for nutrition researchers on Apple Silicon. Your data stays on your Mac. Hybrid embedding and LLM matching available for even higher accuracy.
What it does
Run the entire matching pipeline on your Mac's GPU via MLX. No accounts, no telemetry. Add an optional cloud LLM stage for even more accurate results.
On-Device AI Matching
Embedding model runs natively on MLX with Apple's Metal GPU framework, taking full advantage of unified memory on M-series chips. Your data never leaves your Mac.
Guided Review Workflow
Confirm, reject, or override matches with keyboard-driven review. Auto-advancing guided mode works through items needing attention. Export final results to CSV when you're done.
Built-In Databases
FooDB (9,913 items) and DFG2 (256 items) ship with pre-computed embeddings. Import your own CSV or TSV files too -- embeddings are cached on disk so repeat matches are instant.
Behind the Research
Explore the methods from the research paper with a scrolling showcase and live matching demo. Walk through each stage of the embedding and LLM hybrid pipeline.
How it works
Load your data
Drag in a CSV or TSV with food descriptions. Pick your description column and target database.
Match
Click Match and the embedding model runs on your GPU. Semantic similarity finds the best database entries for each food item.
Review and export
Guided review walks through results. Confirm, reject, or override each match. Export with original columns plus match metadata.
