Skip to content

Settings

matiasdelellis edited this page Jun 4, 2024 · 7 revisions

Settings

The administrator has access to a settings panel where can configure parameters of the analysis and clustering of the faces.

Setting Admin

Sensitivity

The sensitivity determines how different the faces can be to continue to be considered as the same person. For details and advanced information read the documentation about Sensitivity.

Minimum confidence

The minimum confidence determines how reliable must be a face detection to try to group it. Blurred or misaligned faces would have a confidence closer to 0.0, and the best images close to 1.0. For details and advanced information read the documentation about Confidence.

Temporary files

During analysis, temporary files are used to ensure homogeneity between all images. Small images allow a quick analysis, but you can lose the smallest faces of your photos. Large images can improve the results, but the analysis will be slower. See https://github.com/matiasdelellis/facerecognition/wiki/Models#comparison for details.

Note that this setting is limited by the PHP memory limit setting. If you set a small memory value, you cannot analyze larger images.

Hidden settings

There are some preferences that are not exported to the admin panel because they are advanced.

  • default_enabled: This boolean determines if face recognition is enabled by default for all users. By default it is false.
  • clustering_batch_size: This integer defines the maximum batch size for facial clustering. Setting this value improves the speed of clustering and is recommended if you have many faces (more than 25 thousand). The default is -1, which means no batches are used.
  • handle_shared_files: This boolean option add support to analyze shared files. There is discussion about whether we should analyze files shared by other users, but the fundamental reason for disabling this option by default is performance. In order to analyze these images we must obtain a local copy of each file, consuming a large bandwidth.
  • handle_group_files: This boolean option add support to analyze group folders files. Note that in order to analyze these images we must obtain a local copy of each file, consuming a large bandwidth and therefore it is disabled by default.
  • handle_external_files: This boolean option add support to analyze external files. Note that in order to analyze these images we must obtain a local copy of each file, consuming a large bandwidth and therefore it is disabled by default.
  • min_image_size: This numerical option indicates the minimum width or height to analyze any image which is 512 by default.
  • max_image_area: This numerical option overlays the memory configuration, indicating exactly the image size used for the analysis. You can use very small sizes to quickly analyze your entire collection for testing.

Configure it

  • occ config:app:set facerecognition default_enabled --value false
  • occ config:app:set facerecognition clustering_batch_size --value='1000' --type=integer
  • occ config:app:set facerecognition max_image_area --value 786432
  • occ config:app:set facerecognition handle_external_files --value true
  • occ config:app:set facerecognition handle_shared_files --value false
  • occ config:app:set facerecognition min_image_size --value 128

Notes

Note that if you already have the analysis finished and you enable an advanced option, you must force the process to search for new faces.

  • occ user:setting USER_ID facerecognition full_image_scan_done false