flan/winnow
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Update README.md

Browse Source
This commit is contained in:
flan
2026-06-14 13:15:57 -04:00
committed by GitHub
parent 6ec075b9bf
commit 3e030b361e
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.md
+1 -1
View File
@@ -11,7 +11,7 @@
`winnow` pulls photos from your [Immich](https://immich.app) library, selects the most diverse and highest-quality subset using AI embeddings, and delivers them as training data for [Frigate](https://frigate.video)'s face recognition and object classification models.
The best Frigate training data is images you curate manually — photos taken specifically for recognition, in controlled conditions, uploaded directly through Frigate's UI. Winnow is meant to people in your library, not replace manual training. In some cases one has people they would like to recognize that do not occur in detections enough to train a diverse dataset. This is meant to fill that gap.
The best Frigate training data is images you curate manually — photos taken specifically for recognition, in controlled conditions, uploaded directly through Frigate's UI. Winnow is meant to supplement people in your library, not replace manual training. In some cases one has people they would like to recognize that do not occur in detections often enough to train a diverse dataset. This is meant to fill that gap.
> **winnow only touches files it uploaded.** Faces added to Frigate manually through its UI are never deleted, replaced, or modified — not by quality replacement, not by `RESET_PERSON`, not by stale cleanup. Your manually curated images are always the primary dataset; winnow only adds to it.