Pillar 01 · AI Measurement
Per-site camera calibration
Upload labeled video frames, enter what a human counted, and the detector is scored against that ground truth. The resulting class and dwell correction factors are applied to every impression this site reports.
1 · Labeled frame set
2 · Ground truth counts across the uploaded frames
Frames are analysed as objects only — vehicles are counted and classified, never people, faces or plates.
3 · Detector vs ground truth
Run a calibration to see per-class accuracy and the correction factors that will be applied to this site's reported counts and dwell time.
Stored calibration profiles
No calibration profiles saved yet — sign in to view and save profiles.