Match result is half deleted?

What Wildbook are you working in? ACW

What is the entire URL out of the browser, exactly where the error occurred?
March result: Wildbook for Carnivores
Source encounter:

Can you describe what the issue is you’re experiencing?
Previously run match results showed a half blank target image in the match results. I ran a new match and got the same result. Any idea what’s causing this?


Hi Maureen,

This is one of those frustrating EXIF issues. If you zoom out in the match result:

You can see that the animal was correctly detected, but the display of image rotation (as dictated in the EXIF) is wrong. We find images in which EXIF is right and really helps us get all things visibly centered and images where it is wrong and we have to try to workaround it.

It looks like on the Encounter page (different javascript) we are compensating correctly. On the results page, we are confused by the exif rotation. And sometimes the EXIF is right, and sometimes it is wrong (we have cases where humans rotated the image manually but left the EXIF indicating no rotation).

Some good news:

Detection worked fine and found the animal, and it is matchable. We’re just displaying what may or may not be good EXIF incorrectly via the third party seadragon library.

For now, the workaround, which is only for appearances, would be to create a slightly modified JPG without EXIF metadata and upload that image. But otherwise, this is just a display issue.

I am also checking if we already have a ticket for EXIF issue generally.

OK, this issue should be fixed via WB-1696 and deployed for testing on ACW.

We discovered two issues:

  1. This photo was submitted before some EXIF fixes made on ACW, and therefore you should not see this moving forward but may see it again in older data (e.g., some 2020 submitted data) if EXIF rotation is set on an image.

  2. A small adjustment in the results page can help compensate for the rotation better as well. This fix is deployed. And now we can see a good match result at the link above (after we also corrected its orientation by re-running the earlier fix).

I have looked around and also not seen any regression as a result of our fix, but please let me know if you see it (and it’s not a result of older data).