How Rastrum identifies species
Each photo runs through specialized models first. If none is confident enough, an LLM interpreter refines the result. The more models you configure, the more robust your pipeline.
All photos in one deployment share a single GPS. We pull the timestamp per-photo from EXIF.
Trap location (applied to all photos)
Lat/Lng in decimal degrees. Use the same coordinates the camera was deployed at.
Identification will use MegaDetector + SpeciesNet when available. Inference is pending model deployment — observations save now and re-identify on sync once the pipeline is wired.
Drop photos here
or tap to choose files. JPEG/HEIC with embedded GPS work best.
Bulk-set fields
Apply to every included row. Leave a field blank to skip it.
Tap Share, then Add to Home Screen.
Each photo runs through specialized models first. If none is confident enough, an LLM interpreter refines the result. The more models you configure, the more robust your pipeline.