Thanks, @Max.Muehlenhaupt!
Both links load the same set of 12175 encounters for me, but I think that’s just Wildbook dropping the search filter for encounter numbers when I follow a direct search link instead of starting a fresh search.
Just to make sure I’m looking at the correct set of results for each:
Location: Botanischer Garten Bielefeld
Genus and species: Salamandra salamanda
Check box for: Include only encounters for marked individuals that have been sighted at least (2 or 3) times
Assigned researcher: Max
Is this correct? If so, I’ll pass on the details to the ML team to make sure they’re looking at the right collection of data.
Yes, this is correct. The 12175 encounters should be the overall number of encounters with the location ID Botanischer Garten Bielefeld.
1 Like
One small detail I just realized: Since I’m not the only researcher uploading photos for this location ID, using “Assigned researcher: Max” might not result in the overall maximum number of encounters? Feel free to use photos assigned to all researchers (Sean, Saskia, David, Max, …).
If they can post their approval for us to use their data for training in this thread, I can update the search instructions to our ML team.
Hi Anastasia, You can use all data from fire salamander larvae that is assigned to my name. Is that enough for you or would it be better, if I search for something?
And another question. I was just matching some larval pictures. Did you already change something on the matching algorithm? It seems to suggest more wrong matches with a higher score and gives me more options than in the past (see screenshot)
Hi @LauraSchulte
We’ve got what we need for fire salamander juveniles, thanks! We’re still working on training MiewID for adults.
I verified that we haven’t made any changes to Hotspotter. It’s likely that there have been more larvae uploaded, increasing pool of potential match candidates displayed on the match page. The default number of match candidates to show is 12, but Hotspotter will show less than that if the candidate’s match score is 0.
ok. It worked just very reliably in the past and every suggested larva with a higher score than 1 was typically a match. Now I get high scores for no matches
