What is the entire URL out of the browser, exactly where the error occurred?
Can you describe what the issue is you’re experiencing?
I was wondering when we can expect MiewID v4 to be available in Flukebook. It looks like I am still getting results with v3. Additionally, will v4 have the ability to to match left-side photos to right-side candidates (and vice versa) using dorsal fin trailing edge characteristics? In v3 I am only given results for the same side as the query image (i.e. left-side query only produces left-side candidates)
Flukebook was updated to MiewID v4 in January. The way default matching works in all Wildbooks is to find match candidates that share the same location ID, taxonomy, and viewpoints (lefts matched with other lefts such as left front or left back, etc.). The behavior you’re describing is intentional.
To my knowledge, the only species that we don’t enforce viewpoint matching with are elephants in Whiskerbook due to their unique ear positions.
Thank you for your previous answer to this question. Do you know if there is any way to get cross-side matching in Flukebook for Risso’s dolphins? At this point, I am doing all the cross-side matching manually but it would be very helpful for our data analysis if the algorithm could match from the opposite viewpoint (left vs right dorsal fin) so that we can find images of a dolphin from both sides.
We do have the ability to do cross-side matching, which currently has to be set at the species level (a simple configuration change). One of the reasons we have not supported it in the past is the hypothesis that body and fin matching would largely be picking up on the scarring/rake marks (different per side) and less on the fin shape. If the community is largely using fin shape and scarring only secondarily, then it would make sense to turn on cross-viewpoint matching to get fin matches.
@sbent Do you know how the broader photo ID community for Risso’s approaches matching for the species? Is it primarily fin shape?
Thank you for your response. For Risso’s photo ID we primarily rely on scarring but a large portion of these animals have unique fin shapes, specifically very distinguishable notches in the trailing edge of the dorsal fin. In these cases, the notches may become the primary driver of identification. Additionally, using trailing edge characteristics is the primary method for finding a cross side match of an animal. I believe that identification for this species would benefit greatly from utilizing both scarring and fin shape, and allow us to reduce redundancy in our catalogs by avoiding creating duplicate IDs for the same animal from different sides.
Is it possible to have both algorithms for mapping scarring and fin shape/trailing edge characteristics for Risso’s?
Hotspotter uses body matching and MiewID uses bodies and the trailing edge on the dorsal. We can turn off viewpoint matching for the species, but it will turn it off for the bodies too resulting in left viewpoints for bodies when trying to match right viewpoints (for example).
With this in mind, do you still want us to disable viewpoint matching on Risso’s?
If I were to edit an annotation to change it to the opposite viewpoint, would this achieve the same result?
Also, if we do disable the viewpoint matching, would it be a problem to enable it again if it proves to be less helpful in cross side matching than anticipated?
It’s an easy change on our end to disable/enable viewpoint matching, so if you decide it’s not improving your match results, then we can change it back.
Changing the viewpoint manually could work, but with so many extra steps, it would be easier for us to update the matching configuration. I just wanted to be clear that in creating an advantage for dorsal matching, we’d weaken body matching.
You can try it out and if you’re not happy with it, we could switch it back.
@jason Let’s turn off viewpoint matching in Risso’s dolphins (Grampus griseus) in Flukebook so @sbent can test out matchability.