Wildbook connects animal photographs into a history of reappearances

A photograph can answer two different questions: what kind of animal is this, and have we seen this particular animal before? The second question lets a collection of pictures become a history of return visits. It is the difference between recording another sighting and connecting that sighting to an earlier encounter.

Wildbook proposes individual matches for human confirmation. Start with an account arranged through the project’s managers and a species supported by that project. Its first-use guide recommends one photograph of one animal.[1][2]

Connect an encounter

Wildbook’s documentation gives the following first-match route.

1. Choose Submit, then Report an Encounter. Add the photograph, date, location and species; submit it and open View Encounter.
2. Wait for detection boxes, then open Match Results. Compare the proposed matches with the submitted animal.
3. For a confirmed candidate, choose Inspect, then Set to Individual to attach the encounter to that identity.

An important search setting sits behind those results. A supplied location ID automatically restricts candidate matches to that study site; a manual New Match can include several sites. Scores are algorithm-specific, not standardised probabilities.[2]

Keep the observation and the identity separate

Consider an invented example with three photographs, taken in April, June and August. Suppose careful comparison confirms that April and June show the same individual, while August remains unresolved. You now have three encounters, one confirmed identity represented twice, and a remaining identification question. It would be premature to report either three different animals or exactly two.

The unresolved August photograph is still useful. Its date and location remain an observation even while the identity is uncertain. Keep that distinction visible when turning a photo collection into a report: “photographed on three occasions” and “three animals photographed” make different claims.

For a study spanning neighbouring sites, review the search boundary. An empty candidate list from one site cannot establish whether an animal was photographed elsewhere.[2]

Report an error or suggest a correction