I’ve been uploading a slew of field recordings of birds from a recent vacation as .wav files. For one in particular (Brown Creeper), the very high-pitched song and calls are still hard to pick out on the edited .wav file, so I tried to add a screen capture of a spectrograph as further documentation/guidance with the recording. Unfortunately, the uploader is prioritizing the spectrograph over the .wav file in presentation sequence and I can’t seem to reverse that. Any hints on how to get the .wav file to be the primary “image”? or should I just abandon the attempt?
As far as I’ve attempted, it’s not possible. If there is an image, it will always be displayed first.
Just a note that staff have requested users not to upload spectrographs:
You can always post a spectrograph to some other site and link to it as well.
I think I have a slightly different understanding–or a continued misunderstanding–of the way CV is trained:
I understood that a spectrograph should not be the first image in a set, but that it could follow other images. That placement would keep it out of the reach for any training set. But, again, I may be misrepresenting how CV is trained.
In the meantime, I’ll just avoid uploading the specs.
From a prior discussion of what goes into CV
The followup to that always seems to be forgotten or ignored. So let me emphasize again (https://forum.inaturalist.org/t/spectrograph-allowed/31621/4) that this is not a “staff request” - it is @kueda’s personal opinion. It’s very explicitly not policy.
@tiwane Am I misremembering something?–Somewhere I got it in my little brain that only the first image of a set of images in an observation was used in the training sets.
the training set can include photos that are not the first in an observation, but when you ask the computer vision for taxon suggestions, the visual matching for its suggestions will be based only on the first photo in the observation.
@pisum’s correct here.
This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.