Skip to content

Face Detection

Learn how to use AI-powered face detection and manage face regions in your images.

Automatic Face Detection

Running Detection

  1. Click "Detect Faces" button in the Tools section
  2. Wait for AI processing (typically a few seconds)
  3. Detected faces appear with auto-generated names
  4. Review and edit the results

Processing Time

Detection time depends on image size. Larger images take longer to process.

Detection Priority System

When you run face detection, the system intelligently handles existing face tags:

Your manually added tags always take precedence

  • 70% overlap threshold - Faces must overlap 70%+ to be considered the same
  • Existing tags preserved - Your manual tags are never overwritten
  • No conflicts - System automatically handles matches
  • New faces added - Only non-overlapping faces are detected and added

Detection Results

After detection completes:

  • Matched existing tags: Preserved with original names
  • New faces: Added with auto-generated names (Person_1, Person_2, etc.)
  • Status message: Shows number of faces detected
  • Ready to edit: Rename auto-generated names as needed

Detection Accuracy

Works Best With

  • Frontal faces - Direct-facing photos
  • Well-lit images - Good lighting conditions
  • Clear focus - Sharp, in-focus faces
  • Visible features - Eyes, nose, mouth clearly visible
  • Multiple faces - Handles groups well

May Miss

  • Profile views - Side-facing portraits
  • Partially obscured faces - Covered by hands, hair, etc.
  • Very small faces - Faces occupying small portion of image
  • Poor lighting - Dark or heavily shadowed faces
  • Extreme angles - Faces at unusual orientations

Tips for Better Detection

  1. Use high-quality images - Better resolution improves accuracy
  2. Good lighting - Well-lit photos detect better
  3. Frontal angles - Face camera directly when possible
  4. Run detection first - Before adding manual regions
  5. Adjust brightness - If faces are too dark/light

Manual Region Management

Adding Regions Manually

When automatic detection misses a face:

  1. Click "Add Region" button
  2. New red box appears on image
  3. Default name: "Unknown Person"
  4. Position and resize the box
  5. Double-click to rename

Positioning Regions

Move:

  • Click and drag the region box
  • Position over the face area

Resize:

  • Click to select region
  • Drag corner handles to adjust size
  • Fit precisely around face

Naming Regions

  1. Double-click the region box
  2. Type the person's name
  3. Press Enter to save
  4. Name appears in region list

Naming Conventions

  • Use full names: "John Smith" not "John"
  • Be consistent across images
  • Avoid special characters
  • Use spaces between first/last names

Region List Panel

The right sidebar shows all face regions:

Information Displayed

  • Person names
  • Region type (detected vs. manual)
  • Edit button for each region
  • Delete button for each region

Region Actions

From the list:

  • Edit name - Click edit button, type new name
  • Delete region - Click delete button to remove
  • Select region - Click on list item to highlight on image

Region Count

At the top of the panel:

  • Shows total number of regions
  • Updates automatically as you add/remove

Region Colors

Visual indicators show region status:

  • Green - Successfully saved/matched
  • Red - New/unsaved region
  • Yellow - Currently selected for editing

Editing Workflow

  1. Upload image with existing metadata (if any)
  2. Run face detection to automatically find faces
  3. Review detected faces - Check for accuracy
  4. Rename auto-generated names - Replace Person_1, Person_2, etc.
  5. Add missed faces - Manually add any faces detection missed
  6. Delete false positives - Remove incorrectly detected regions
  7. Save your work - Use Save Image or Save & Exit

Priority Handling Example

Before detection: - You manually tagged: "John Smith" at position (100, 100)

After running detection: - AI finds face at (105, 105) - 75% overlap with John Smith - Result: John Smith tag preserved, AI detection ignored (overlap > 70%)

If AI finds face at (500, 500): - No overlap with existing tags - Result: New face added as "Person_1"

Clearing Selection

To deselect a region:

  • Click "Clear Selection" button
  • Or click on empty area of canvas
  • Or select a different region

Keyboard and Mouse

Canvas Navigation

  • Pan - Click and drag image
  • Zoom - Mouse wheel or zoom buttons
  • Select - Click on region box

Region Editing

  • Move - Drag region box
  • Resize - Drag corner handles
  • Rename - Double-click region box

Best Practices

For Accurate Detection

  1. Start with good quality images
  2. Run detection on unedited photos when possible
  3. Use consistent naming across your photo collection
  4. Add manual regions for missed faces
  5. Review all auto-generated names

For Efficient Workflow

  1. Run detection first (before manual work)
  2. Batch rename similar auto-generated names
  3. Delete obvious false positives immediately
  4. Save frequently (auto-save runs every 30 seconds)
  5. Use meaningful, searchable names

Name Consistency

  • Good: "John Smith", "Mary Johnson", "Robert Brown"
  • Bad: "john", "Mary J", "Bob"

Consistent naming helps when:

  • Searching for people across multiple images
  • Organizing your photo collection
  • Exporting to other photo management tools

Troubleshooting

Faces Not Detected

Check:

  • Face orientation (frontal works best)
  • Image lighting (not too dark/bright)
  • Face size (not too small in frame)
  • Image quality (sharp, in-focus)

Try:

  • Adjust image brightness/contrast before upload
  • Add manual regions for missed faces
  • Use higher resolution version of image

Too Many False Positives

  • Solution: Delete unwanted regions from region list
  • Common false positives: Patterns, objects, reflections
  • Not a problem - easily removed

Can't Select Region

Check:

  • Click directly on region box (not just near it)
  • Try zooming in for better precision
  • Clear selection and try again

Detection Takes Too Long

  • Large images take longer
  • Wait for completion (don't navigate away)
  • Consider using smaller image if extremely slow

Next Steps