Every report you generate is saved here so you can compare runs — how detection counts move
with region, threshold, and image. Faces/plates are pulled from Redact's labels output.
| When | Mode | Risk | Imgs | Detections | Recall | Images | |
|---|---|---|---|---|---|---|---|
| No runs yet. Generate one → | |||||||
Tip: run the same image twice with different threshold sweeps, then compare the detection counts here to talk through the precision/recall trade-off in the interview.