You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We have done experiments with per-clade models (the big-vs-small experiment we did a few years ago), but we haven't explored making regional models and/or regional data packs. We could transfer learn the model from the 2.0 feature extractor.
For example, a user could use the app to download a regional package for North America that includes a model, a realm database, and a bundle of images. We could use that to pre-seed the taxonomy database and an image cache, so that even while offline the user would get locally relevant suggestions with common names and thumbnails.
The text was updated successfully, but these errors were encountered:
We have done experiments with per-clade models (the big-vs-small experiment we did a few years ago), but we haven't explored making regional models and/or regional data packs. We could transfer learn the model from the 2.0 feature extractor.
For example, a user could use the app to download a regional package for North America that includes a model, a realm database, and a bundle of images. We could use that to pre-seed the taxonomy database and an image cache, so that even while offline the user would get locally relevant suggestions with common names and thumbnails.
The text was updated successfully, but these errors were encountered: