Rebuild Database Indexes

Rebuild is used to repair or regenerate feature data for files that are already in the database. It is different from refresh, which mainly syncs file changes.

Steps

1. Open the rebuild database dialog.
2. Choose the database area or feature data that needs repair.
3. Select the strategies you want to rebuild.
4. Start the task and watch progress.
5. After completion, try the related search again.

Which Database Should You Choose?

• File database: use it when file records, metadata, or file list behavior looks abnormal.
• Image database: use it for image search related capabilities, including global, background-removed, color, and local strategies.
• Pose database: use it when pose search data needs to be regenerated.
• Semantic database: use it when text-prompt search is unavailable or abnormal.
• Face database: use it when face search or face albums need to be regenerated.

Choosing Image Strategies

• Global: rebuild normal reverse image search.
• Global with background removal: useful for clear subjects with distracting backgrounds.
• Global color: rebuild color-aware search data.
• Local keypoints: rebuild local-region matching data.

Difference from Refresh

Refresh checks what changed in folders and updates records. Rebuild focuses on existing records and recreates search indexes or feature data. Use rebuild when counts do not match, a strategy is missing, or search behavior is clearly wrong.

Common Cases

• If a search strategy shows no results even though files exist, rebuild that strategy.
• If face albums are incomplete, rebuild face data.
• If duplicate detection was enabled later, rebuild duplicate-related data before checking duplicates.