Why does the number of "candidates" vary between matches within a project?

Flukebook / Mac Os High Sierra 10.13.6 / Firefox, Chrome and Opera / Admin

We have some questions regarding the matching process within a project framework.

In fact, the number of “candidates" (see the red square in the screenshot below) differs between matchs and differs from the number of encounters of a given project. For instance, for a project containing 3729 encounters, the number of candidates against which the image is matched can vary between 3420 and 4450. This seems ackward, since within a given project, images are supposed to be matched again all images of that project, so the number of candidates against which each image is matched should not change. As it si not the case, could you please explain what the number of “candidates” refers to exactly and why it is changing depending on the image being matched ? Can we be sure that all encounters (i.e. all encounters/images included in a project) are considered for each matching process ?

Thanks for your answers,

Paul and Violaine

Hello @paul_lallement - I work on a wildbook for terrestrial species, so a little different from yours but some of the factors that cause the # of candidates to vary are: location ID filter(s) in the match criteria, as well as viewpoints - the system will always try to match same/similar viewpoints to each other and will exclude opposing viewpoints (eg. tops matched only against other tops, left against only left viewpoints, etc). I hope that helps!

best regards
Maureen

Hello Maureen,

Thank you for your prompt and useful answer.

According to the Wild Me Docs (c.f. “Start match“ section from “Understanding projects“ part : Projects | Wild Me Docs), within a project, matching is carried out only among encounters of the current project. Once a project has been created, there is no option to match against a specific Location ID, therefore Location ID should not have an effect on the number of candidates in the framework of a project.

Thus “viewpoint” (rather than LocationID) might affect the number of candidates (thanks for the info, we didn’t think of that). In our dataset, most annotations were done automatically during the bulk import. Therefore, the viewpoint must have been defined automatically as well. However, it seems that the info re. viewpoint is not longer displayed once the images have been annotated. Do you know where we can check the viewpoint of a given annotation? We are working on humpback whale’s fluke and most images are taken from the back, so we need to make sure that all image’s viewpoints have been defined as “back”.

Thank you in advance for your answers.

Best regards,

Paul and Violaine