Probably best not to invent observations like this. You can try out plantnet itself as it can deal with plants in most parts of the world from its website, we are only using it on ispot for UK at moment. If you did want to see Veratrum in UK then I am sure it will be at a number of botanical gardens and similar. e.g. British Isles garden and greenhouse plants although don’t have access to my database at moment to find exact coordinates of this image.
Ok. I won’t do that.
What’s wrong with humans?
Comments
I wonder why you are not accepting that we should test AI to the full
Is my test of the artificial flower to be frowned upon? You haven’t said
I have a photo of two very similar wild flowers side by side and Plant@nets response. Should I share the response here?
Yes it is ok to talk about issues to do with AI here but what I was suggesting was to do any ‘tests’ of the AI on the AI site itself and not in an Ispot observation which might be turned into a false record.
here…Impressive?
Oh yes I agree not to use iSpot and post negative results. AI itself will do a few of those.
I will turn the above into a glowing support of the AI App.
Not everyone will know it can be PC based as well as Cell-phone
The results we see in iSpot trials are, of course, PC based
Re the Working Horse observation. I cannot recognise what the image is in the AI contribution of Bos.
.
I think it’s a fossil Jo
one like it for £45 https://www.hemswell-antiques.com/antiques/antiquities/mammal-foot-bone-fossil-76300.html?gclid=EAIaIQobChMIxebLkbm8-QIV8mDmCh3nPQS7EAQYASABEgJRJvD_BwE (preview removed)
OK I am impressed.
What is the purpose of running this IN iSpot?
Will users change their ID as a result?
Will knowledgeables respond to Plant@Net or the user?
Will we go off-tangent rather that adding advice about how to ID or form an Observation or use the Drop-down menu?
Will Experts stop interacting because they cannot discuss anything with AI?
.
Using my phone is a waste of time - it cost me £1,49 to do a Euprahasia on my PayAsYouGo
Even when my Cell-phone was working quite well, I walked in and out of signal waiting for a result and eventually failed.
I am quite happy testing this but I’d like to see two boxes in these
https://www.ispotnature.org/communities/uk-and-ireland/view/observation/848881/ Preview removed
1…Please grade the result 1(poor) to 10 (perfect)
2 If it is less than ‘reasonable’ please add a link to the Forum item here iForum LIVE!: Why are we integrating FASTCAT-Cloud and the Plantnet API into iSpot?
1 Like
Re Reminder
After seeing the AI comment, I put the lead photo from this obs into Plantnet on my phone, it said 99% Cynosurus cristatus! (I was in W Europe flora.) The second suggestion it gave is tropical/subtropical/warm temperate acc to Wikipedia. Why offer three here on iSpot when 99% certainty offered by Plantnet app? Or does it always give three anyway, so where runners-up are unlikely, nonetheless they are shown? Does it take location into account? (Sorry, tried to read through all the points in this thread but failed to get my head round quite a bit of it)
Also takes up a lot of space if people are using a small screen to view iSpot so human comments may not be so readily seen, even with the helpful Comments Present banner.
1 Like
There is an issue with Eyebright (Euphrasia) in that it proposes species.
The same probably applies to Dandelion
Most people who read this will understand why it is an issue
I agree with Maggs49 about location filtering. A recent test suggested a species not recorded in the UK.
I have included that test in an Observation
In the Euphrasia example it gives percentages.
I think that most of us taking part in this forum topic will recognise that such low percentages indicate that the Genus is probably as far as PN reasonably can go.
.
Whether beginners will know that, who knows. I guess we can offer them guidance in such circumstances.
.
A sort of p.s…… the percentages don’t appear on the PN ispot is using. I think they could be useful.
This is probably the first example of suggested ids for a seaweed. Interesting suggestions albeit for plants not seaweeds. We need a human!
I am human and I have warbled there…
It wouldn’t do any better with mosses, liverworts and hornworts either - I’ve just checked Pl@ntNet’s corpus, and it’s vascular plants only.
Suspect it does not do these other groups yet as there are not enough images of the species in these groups to train it on. On the other hand there might be enough images if they were somehow amalgamated. If humans can recognise these general groups then there must be something that the AI can work with.
Oversight? @miked @Janice_A
did you know we are not being flagged (Changes Tracker) when P@N adds a ‘comment’. That should be corrected, as I suspect many users have no idea this is going on.
I invented a tag today - any use? If those who get a comment from P@N added it, we would all see them
https://www.ispotnature.org/communities/uk-and-ireland/view/observation/849154/ preview removed (by me)
Is there still no way that the Activity Tracker can be updated here?
Pl@ntNet | User | iSpot Nature
so that we can SEE the 155 suggestions
@Chris_Valentine have tagged chris in to this one as there are two parts. First do we actually want to put it into the users track changes and second it is possible.
The answer to the second part might be no as the comments are put in a different way to normal comments so may need some heavy duty coding to change the track changes to allow this, Chris should be able to answer this.
As to the first part, the answer might again be no at least once people understand the system as they will know that the observation is going to get a suggestion from plantnet automatically anyway after a certain time. On the other hand it might be useful to have these comments come up in the tracker at least in the early days of the implementation to remind people that they will have at least one suggestion of a name even if noone else has responded.
plant | Observation | UK and Ireland | iSpot Nature is an interesting observation to keep an eye on. Tried SEEK then Plantnet (manually outside of ispot) and I think plantnet got it right. Should also have the Plantnet automated AI comment soon to see what that says.