Quality
Review queue & confidence
AI accelerates labeling. People keep quality under control.
Not every capture is equally clear. Night shots, partial animals, or unusual angles can lower the model’s confidence. When a detection falls below the threshold (in the platform, typically around 70%), it is held in the review queue until someone confirms or corrects the species.
- Confirm as-is — accept the AI suggestion when it looks right.
- Correct the species — pick the right label; the correction is stored.
- Bulk actions — clear many items during busy seasons.
Reviewed detections join the same charts, maps, gallery, and exports as automatic high-confidence results — with a clear quality trail.
