Skip to content

Editing Encounters

  • Where the encounter edit experience lives
  • Which fields are editable and which require admin or professional rights
  • How edits interact with annotations, ML, and the encounter’s status
  • The split flow for partitioning one encounter into two

Editing is a right-side slide-over drawer, not a separate page. From any encounter you have permission to modify, open the encounter detail view and trigger Edit from the action panel. The drawer slides in from the right with the encounter pre-loaded; the gallery and metadata behind it remain visible so you keep your context.

Image management — adding, deleting, finding duplicates, setting fin sides — is a separate surface. See Gallery Management. The edit drawer is for metadata; the gallery view is for photos.

These mirror the creation form and are available to anyone who can edit the encounter:

  • Date — full date picker
  • Location — opens the location picker dialog; you can reassign to a different existing location or create a new one
  • Description — encounter notes shown publicly with the encounter (10,000-character limit)
  • Behaviors — multi-select from the population’s behavior list
  • Completeness — Complete / Incomplete / Unknown radio
  • Predation event — flag plus optional prey targets (species, count, location, time)
  • License — change the encounter’s data-limitation license

These render only when your role grants edit access beyond the standard owner level:

  • Submission user — reassign the encounter to a different photographer. Optionally check Inform user to email the new submitter when the change is saved. Useful when an encounter was originally submitted on someone’s behalf and the real photographer’s account becomes available later.
  • Organizations — multi-select; controls who can see the encounter when the license is organisation-restricted.
  • Analyst notes — separate field from Description; intended for internal commentary that shouldn’t appear in public-facing views. 10,000-character limit.
  • Population transfer — move the encounter to another population entirely. This fires a backend queue job after a warning dialog; the move runs asynchronously so the UI doesn’t block.
CapabilityRequired role
Edit standard fieldsEncounter owner, Professional, or Administrator
Reassign submission userProfessional or Administrator
Change organizationsProfessional or Administrator
Move to another populationProfessional or Administrator
Edit gallery / add or remove imagesSee Gallery Management

The drawer hides controls you don’t have access to rather than showing them disabled, so the form you see is your effective permission set.

How edits interact with ML and annotations

Section titled “How edits interact with ML and annotations”

Metadata edits do not automatically re-run the ML pipeline. Changing the date or location doesn’t invalidate any existing annotations, and the model isn’t asked to re-evaluate the images.

Two related actions sit nearby and do affect ML state:

  • Setting status to Finished triggers cleanupPredictions — extraneous unconfirmed model predictions are removed once the encounter is considered done. This is a one-way transition tied to the status change, not to general edits.
  • Re-queue annotations is available on the encounter action panel below the drawer (Re-queue unfinished and Re-queue all). These force the pipeline to re-process — useful after gallery changes or model upgrades. They are not in the edit drawer because they’re a workflow action, not a metadata edit.

If an encounter was merged or split, the Re-associate action reconciles its annotations against the new individual layout. This is also outside the edit drawer.

The same drawer reopens in split mode when you start a split from the gallery. Split mode lets you select a subset of photos and carve them into a new encounter — useful when a single submission accidentally bundled two distinct observations (e.g. two non-interacting groups photographed on the same day).

The flow:

  1. From the gallery, choose Split encounter.
  2. The drawer opens with the parent encounter’s metadata pre-filled and a photo picker.
  3. Pick the photos that belong to the new (child) encounter.
  4. Adjust any metadata that should differ for the child (date, location, behaviors).
  5. Save. The child encounter is created; the picked photos and any of their confirmed annotations move with it.

The parent encounter keeps its remaining photos and metadata. ML annotations on photos that moved travel with their images.

  • Submission-user change with “Inform user” sends an email notification to the newly assigned photographer.
  • Status change to Finished runs prediction cleanup (described above) and fires the encounter-status-change notification.
  • Population transfer queues a backend migration job; you’ll see the encounter disappear from the current population’s lists once the job completes.
  • Standard edits are persisted via the encounter update API and logged for audit.

The drawer also surfaces a Delete action for administrators, in a clearly-marked danger zone. Deleting an encounter is permanent and removes its annotations from the population’s sighting record.