Hunterx

Accuracy and confidence guide

Upload
Confidence scoring

Every `.hunterx` file should explain the strength of its measurements.

Explainable
0.00-0.29

Very low

Intake/debug only

0.30-0.49

Low

Rough single-image estimate

0.50-0.74

Medium

Draft quote after review

0.75-0.89

High

Normal orthomosaic quote workflow

0.90-1.00

Survey

RTK/GCP/checkpoint-supported workflow

Drone recommendation tiers

The drone changes the ceiling. The capture quality and reconstruction decide the score.

Controlled estimate

DJI Mini 5 Pro, Mini 4 Pro, Air 3S

Confidence

Low to high, depending on capture quality and reconstruction

SurveyingMode

Do not assume SurveyingMode = 1

Product stance

Best for our V1 controlled capture set and quote estimates. These drones can provide useful image metadata, but they are not the same as RTK mapping platforms.

Mapping workhorse

DJI Mavic 3 Enterprise / Mavic 3E

Confidence

Medium to high, higher with RTK module and clean reconstruction

SurveyingMode

Expected in enterprise mapping metadata

Product stance

A stronger fit for mapping missions because Mavic 3E supports a mechanical shutter, JPEG/DNG stills, and an RTK module.

Survey-supported

DJI Phantom 4 RTK, Matrice 4E, Matrice + Zenmuse P1 class

Confidence

High to survey when RTK/GCP/checkpoints pass

SurveyingMode

Expected in mapping/survey workflows

Product stance

Best fit when customers need the strongest evidence chain. The confidence score should still show GCP/checkpoint and reconstruction evidence.

SurveyingMode

DJI mapping metadata signal.

Value 0

Warning

Not recommended for mapping accuracy. Hunterx should reduce confidence and explain why.

Value 1

Positive

Recommended for mapping accuracy. Hunterx should still verify overlap, RTK/GPS quality, blur, exposure, and reconstruction output.

Evidence signals

Signals that move the score up or down.

Single DNG/JPEG

Low baseline

Full overlap set

High potential

Nadir gimbal

Positive signal

RTK fixed

Survey signal
Language standard

How Hunterx should talk about accuracy.

Single-image estimates can be useful, but they should be labeled low confidence unless validated by a reference distance, ground control, DEM, or overlapping reconstruction.

Quote-ready and survey-supported are different product claims. The `.hunterx` file should carry enough evidence for both the UI and exported PDFs to make that difference clear.