AI processing

Every uploaded photo is processed automatically in the background at PhotoGala: the NSFW content filter detects inappropriate material and routes it to moderation, and face recognition finds and groups people so your guests instantly find their own photos via people search. The AI processing dashboard shows you, the host, how far that processing has come, where it is stuck, and which feature is currently active.

Phone

Computer

The AI processing dashboard: overall progress, feature status, distributions.

Note

The dashboard is monitoring only. AI features are turned on and off in Settings (the gear icons link straight there). Only the host (owner role) manages this dashboard. Moderators just see the consequences: NSFW uploads land in their moderation list.

How to reach this area

There are two ways into the AI dashboard.

1. From Event statistics. Open the admin area and go to Event statistics. At the very bottom you find the AI Processing Status card with the subtitle “Monitor NSFW filtering and face detection progress.” Click View there. This takes you to /admin/ai-processing-stats.

   Phone

   

   Computer

   

   Entry via Event statistics: the AI Processing Status card at the bottom, button View.

2. From the admin search. Type “AI Processing” into the admin area’s global search (the keywords nsfw, faces, detection, processing also match). The result takes you straight to the same page.

3. At the top of the page a breadcrumb reads Event Statistics › AI Processing. Use Event Statistics to return to the statistics overview at any time.

Caution

If your event plan includes no AI feature and your event is branded (managed through a partner/brand), the page redirects to the dashboard automatically — branded events cannot self-upgrade. For non-branded events the statistics card instead shows an Unlock AI processing block with NSFW Filter and Face Recognition buttons that lead to the pricing page.

The dashboard header

At the very top sits the AI Processing Dashboard card with the subtitle “Monitor automatic content filtering and face detection”. It summarises the overall state.

• Refresh: The top-right Refresh button reloads the statistics; while loading it reads Refreshing… and the arrow spins. Data stays “fresh” (cached) for about 30 seconds — after a big upload wave, refresh once.
• Overall Progress: The large percentage shows overall progress; it turns green at 100%.
• Progress bar: Below it reads “N of M images processed” with “N remaining” on the right. At 100% the All Complete chip with a checkmark appears.

Tip

Overall progress is averaged only across the enabled features. If no AI feature is active, the bar area reads “No AI features enabled. Enable features in settings to start processing.” and progress counts as 100% — there is simply nothing to process.

The “AI Features Configuration” card

Right below the header sits the AI Features Configuration card (“Current AI processing settings - click cog icons to configure”). It is your status and quick-access panel, with one row per feature.

• NSFW Content Filter — “Automatically detects and filters inappropriate content”. On the right a status chip Enabled or Disabled and a gear icon. The gear jumps into the Privacy & Moderation settings and highlights the NSFW filter toggle there.
• Face Recognition — “Detects and groups faces in your photos”. Chip Enabled/Disabled plus a gear. The gear jumps into the Feature settings and highlights the face recognition toggle.

Note

Locked state: If a feature is not in your plan (and your event is not branded), a large coloured card replaces the row — Unlock NSFW Filter or Unlock Face Recognition — each with a button that leads to the pricing page. If both features are in your plan but switched off, a yellow No AI features enabled notice also appears, with “Enable AI features in settings to start processing your media.”

Feature status cards

For each enabled feature a dedicated status card appears (side by side when both are on).

• Header: icon, title (NSFW Filter or Face Recognition) and the subline “N of M done”.
• Percentage: a large % value on the right; at 100% the Complete chip with a checkmark replaces it. The border colour signals state: green = done, red = there are errors.
• Progress bar in green/orange/blue.
• Status row (only what applies):
  • Yellow dot “N pending” — waiting to be processed.
  • Pulsing dot “N processing” — running right now.
  • Red warning icon “N failed” — clickable, opens the failed-media dialog (tooltip “Click to view failed media”).
  • Grey dot “N excluded” — media uploaded before the feature was enabled, so they are not (any longer) processed.
  • Green All processed at 100%.

Tip

Read pending and processing as normal live progress — they clear on their own. The only value that requires action from you is failed.

Distribution cards

Below the status cards, distribution cards show the top-3 categories per feature with bars and percentages. If there is no data yet, they read “No data processed yet”.

• NSFW Distribution (only when the NSFW filter is on): Safe / Questionable / Unsafe / Not Processed.
• Face Detection (when face recognition is on): Faces Detected / No Faces / Error / Not Processed.
• Face Recognition: Grouped / Embeddings Extracted / Not Processed / Error.

Note

Grouped means the faces of one person have been merged together — this is exactly what powers people search for your guests. Embeddings Extracted is a technical intermediate step before grouping.

Handling failed media

The red “N failed” button on a status card opens the Failed AI Processing dialog (subtitle “NSFW Filter” or “Face Recognition” with the count).

1. Click “N failed” on the affected feature card. The dialog lists every failed item with a thumbnail, Media ID (shortened), the failure time, a “X/3 retries” chip (max 3 attempts), a Max retries chip if applicable, and a red Error: box with the error message.

2. Select the media: individually via their checkbox, or all of them via Select All at the top (with a “selected/total” counter). The footer shows “N selected” on the left.

3. Retry Processing (blue button): re-queues the selection for processing. On success a toast reads “Queued N media for reprocessing”. Media that has reached the 3/3 limit is skipped — a toast then reports “N media skipped (max retries reached)”.

4. Ignore (red outline button with a crossed-out eye): marks the selection as ignored; it disappears from the list. Use this for media that can never be processed.

Caution

An item that has hit the limit (Max retries, 3/3) can no longer be reprocessed — only Ignore helps here. If you select nothing, or only non-retryable media, a notice tells you. After Retry or Ignore the list refreshes automatically.

Saving & applying

The AI dashboard saves nothing — it is pure monitoring plus the Refresh, Retry and Ignore actions. Actually turning features on or off happens in Settings (via the gear links). After you toggle and save there, processing starts; pull the state into the dashboard with Refresh. Retry and Ignore refresh the statistics automatically.

Plan & unlocking

• The NSFW filter and face recognition are each tied to your event plan. If a feature is missing, you see the unlock block with an upgrade button to the pricing page.
• Branded events show no upgrade block (they cannot self-upgrade); with no AI feature the page redirects to the dashboard.
• Even with the right plan, the feature still has to be enabled in Settings for the status and distribution cards to appear.
• Enabling face recognition requires explicit consent (GDPR/biometric data). See the Feature settings guide for the details.

Troubleshooting

• I see no feature cards. The feature is in your plan but switched off in Settings — open it via the gear and enable it.
• Everything shows 0 / “Not Processed”. No new uploads processed yet, or the feature was just enabled. Excluded means: uploaded before the feature was on.
• The failed count stays. Try Retry; at 3/3 only Ignore helps.
• Progress is stuck. Press Refresh (30 s cache); processing/pending are normal intermediate states.
• The page redirects to the dashboard. A branded event with no AI feature in its plan.

Best practices

• Enable the NSFW filter before you open public or guest uploads, so questionable material flows automatically into moderation.
• Enable face recognition early, so guests instantly find their own photos via people search — and mind consent (privacy).
• After big upload waves, press Refresh once and check the failed list.

Related help

• Privacy & Moderation (settings) — the NSFW toggle, threshold and auto-reject.
• Feature settings — enabling face recognition including consent.
• Moderation — where NSFW uploads go.
• Event statistics — the entry point into the dashboard.
• People search (guest view) and privacy — what face recognition does for guests.