Sharkbook - worsening results for spot matches

Bug Reports
,
1

Sharkbook

I am experiencing a worsening set of results of spot matching within the Carcharias taurus species. I never used to rely much on i3S algorithm, but recently the modified groth algorithm has been really poor. The other day I reviewed ~150 sharks before I got to the correct one.

Here are some recent example where there were no results from Modified Groth, and then some matches in the i3S algorithm (not always top of the list either):

Sharkbook: Wildbook for Sharks - On this example, the only match it found was the duplicate image. More images were available and yet it was not finding them. I looked through all 15 of the modified groth matches, and I looked through all 91 of the i3S matches, and the only match it found was the duplicate image.

  • Sharkbook: Wildbook for Sharks - On this example, I am really surprised as the correct shark (GNS-0191) has so many images that could have helped the AI find the correct shark, and yet the modified groth returned zero results. When we switched to i3S it found the match – but usually the groth algorithm was reliable?

I saw a similar post about nursehounds and wondered if it was related?

2

I answered this over here, but the short answer is that spot mapping for Carcharias taurus will be replaced with MiewID by next week.

We’ve been focusing our resources on improving MiewID. For context, when we see reports of spot mapping not working correctly, our incentive is to get MiewID ready faster instead of trying to fix a tool we’re actively trying to replace.

3

Quick update: New ETA for MiewID for Carcharias taurus in Sharkbook is next week. We’re aiming for Monday, but definitely sometime next week.

1 Like