I use both the iNaturalist app and the Seek app (which I use mainly for getting a quick id on weeds in my garden) but I keep the two separate. I have noticed that sometimes, where the iNaturalist app will only suggest to the genus level, Seek will readily identify to the species level using the same image. I had assumed that they would both use the same database. Curious as to why this is the case. I confess I will happily manually enter Seek’s more specific suggestion when submitting an observation if iNaturalist doesn’t.
I use the Seek ID as a point of comparison, as CV isn’t always right, in either app.
I use Seek for its faster (sometimes) answers in the field and I like what the Seek camera does with my clip on lens. I rarely upload observations from the field.
I have no idea why they aren’t the same.
Seek is using an older version of the CV (unless you’re using the offline version in the new iPhone app), and I think it is meant to give a species ID. So I would consider the iNat app generally to be more accurate.
Yes, Seek and the new app in offline mode use a smaller model that can fit on a phone. This is updated less frequently. The apps/webpage use the full, current model. If you have access to suggestions from both, the iNat app will generally be more useful/reliable as it includes more taxa.
Do they use geographic boundaries or time boundaries ? And the (offline) Seek model is differnt from the offline model of the new App?
The most recent versions of Seek and iNaturalist use the geomodel.
Both are using 2.13.
I’ve found that Seek often is far less specific than are iNaturalist’s suggested identifications. More than once after Seek has identified a plant as a “dicot,” I’ll switch to iNaturalist for a suggested identification that includes both genus and species.
Because Seek can only give you one ID suggestion, its confidence threshold is set pretty high.