administrator/frigate
1
0
Fork 0
mirror of https://github.com/blakeblackshear/frigate.git synced 2026-03-14 20:37:34 +03:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

Simplified the Frame Selection section down to a single paragraph.

Browse Source
This commit is contained in:
Kai Curry 2026-02-09 16:54:40 -01:00
parent d48a23292a
commit a6a5b6772d
1 changed files with 1 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
docs/docs/configuration/snapshots.md
View File

@ -13,13 +13,7 @@ Snapshots sent via MQTT are configured in the [config file](/configuration) unde
## Frame Selection
Frigate does not save every frame — it picks a single "best" frame for each tracked object and uses it for both the snapshot and clean copy. As the object is tracked across frames, Frigate continuously evaluates whether the current frame is better than the previous best. A frame is considered better if:
- It has a **face** attribute (for `person` objects) or **license_plate** attribute (for `car`/`motorcycle` objects) with a larger bounding box area than the current best. Once the current best frame has one of these attributes, it can only be replaced by a frame with an even larger one — score and area improvements alone won't override it.
- The object's detection **score** is more than 5% higher
- The object's detected **area** is more than 10% larger
All scores and areas above refer to the parent object (e.g. `person`, `car`), not the attribute. Frames where the object's bounding box touches the edge of the frame are deprioritized. Image quality (e.g. blurriness) is **not** considered during frame selection — the [`blur_confidence_filter`](/configuration/face_recognition) setting only affects face recognition confidence, not snapshot selection. The snapshot is written to disk once at the end of the event using whichever frame was determined to be the "best".
Frigate does not save every frame — it picks a single "best" frame for each tracked object and uses it for both the snapshot and clean copy. As the object is tracked across frames, Frigate continuously evaluates whether the current frame is better than the previous best based on detection confidence, object size, and the presence of key attributes like faces or license plates. Frames where the object touches the edge of the frame are deprioritized. The snapshot is written to disk once tracking ends using whichever frame was determined to be the best.
MQTT snapshots are published more frequently — each time a better thumbnail frame is found during tracking, or when the current best image is older than `best_image_timeout` (default: 60s). These use their own annotation settings configured under `cameras -> your_camera -> mqtt`.