When Faces Are Not Being Recognized
Flock's photo attendance uses AI to identify members in a group photo and automatically mark them as present. It works well in most situations, but there are times when the AI struggles to match faces. Here is what causes recognition issues and how to fix them.
Common Causes and Solutions
Poor Lighting
The problem: Dim rooms, strong backlighting (like a window behind the group), or harsh shadows make it difficult for the AI to distinguish facial features.
The fix:
- Face the light source. If there is a window, have the group face toward it rather than away from it. Natural light on faces is best.
- Turn on the lights. In a living room or meeting space, flip on overhead lights and lamps before taking the photo. More light means better recognition.
- Avoid silhouettes. If the group is sitting in front of a bright window or screen, the AI sees dark shapes instead of faces. Move the group or close the blinds.
Members Without Profile Photos
The problem: The AI matches faces in the group photo against each member's profile photo. If a member has not uploaded a profile photo, the AI has no reference to compare against.
The fix:
- Encourage everyone to add a profile photo. Mention it at your next meeting: "Take a moment to add a profile photo in the Flock app — it helps our attendance photo work better."
- Use a clear, well-lit headshot. Profile photos work best when the face is clearly visible and well-lit. A casual selfie works perfectly.
Obscured Faces
The problem: People looking down at their phones, turned to the side, wearing hats or sunglasses, or partially hidden behind someone else.
The fix:
- Ask everyone to look at the camera. A quick "Everyone look up!" before snapping the photo makes a big difference.
- Take the photo from the front. A straight-on shot works much better than an angle.
- Hats and sunglasses. If possible, ask people to briefly remove hats and sunglasses for the photo. The AI relies on facial features that these can obscure.
Large Groups
The problem: In very large groups, people in the back of the photo may appear too small for the AI to reliably identify.
The fix:
- Take multiple photos. Split a large group into two or three sections and take a separate photo of each.
- Move closer. If possible, get closer to the group so faces are larger in the frame.
- Use manual check-in for stragglers. Take a photo of the main group and manually add anyone who was hard to capture.
Photo Quality
The problem: Blurry photos, motion blur from a shaky hand, or low-resolution images reduce the AI's ability to identify faces.
The fix:
- Hold steady. Take a moment to stabilize your phone before tapping the shutter.
- Make sure faces are in focus. Tap on a face on your screen to set the focus point before taking the photo.
- Avoid zooming. Digital zoom reduces quality. Move closer to the group instead.
New Members
The problem: A member just joined the group and has not been matched by the AI before. First-time recognition can sometimes be less reliable.
The fix:
- Make sure their profile photo is set. This is the reference the AI uses.
- Identify them manually the first time. When the AI shows an "unknown face," select the person from your member list. Future recognition should improve.
What to Do With Unknown Faces
When the AI cannot identify a face, it shows you the face and gives you options:
| Option | When to Use |
|---|---|
| Identify as existing member | You recognize them — tap their name from the list |
| Mark as visitor | They are a first-time guest |
| Ignore | The face belongs to someone not in the group (a barista in the background, etc.) |
These manual corrections help the AI learn. Over time, recognition accuracy improves as the system builds a better understanding of each member's appearance.
When to Use Manual Mode Instead
Sometimes photo attendance is simply not the right tool for the moment. Switch to manual check-in when:
- The meeting is virtual. Zoom calls and phone conferences do not lend themselves to group photos.
- You are meeting outdoors in challenging light. Campfires, sunset meetings, or low-light outdoor settings.
- Only a few people showed up. For a group of three or four, tapping their names is faster than taking and processing a photo.
- You are recording attendance after the meeting. If you forgot to take a photo during the meeting, just manually mark who was there from memory.
Manual and photo attendance both feed into the same analytics. Your attendance data, trends, at-risk alerts, and group health scores work the same way regardless of which method you use.
Tips for Better Photo Attendance
Take the photo early in the meeting. People are more settled, less likely to be looking at their phones, and everyone is in the same spot.
Make it a fun ritual. "Group photo time!" turns attendance tracking into a positive moment instead of an administrative task.
Review the results before saving. Take a few seconds to review the AI's matches. Catch any misidentifications or missed faces before you confirm.
Combine methods when needed. Take a group photo for most of the room, then manually add anyone who was late, called in remotely, or was not in the frame.
Related Help Articles
- How to Take Attendance for Your Church Group Meeting — Overview of all attendance methods
- Manual Attendance: When Photo Mode Isn't an Option — Using manual check-in
- Attendance Trends: Spot Patterns Before They Become Problems — Understanding your data