Temporal Change Detection Rate (TCHDR)

GMI-MIN-014MiningIntegration-Required

Definition

Proportion of progressive geotechnical changes that the temporal comparison algorithm correctly identifies across sequential inspection flights.

The Measurement Gap

The Measurement Gap

Drone inspection programs report survey completion rates and anomaly counts as their primary performance metrics. A program that surveys the entire pit wall twice a month with zero missed areas looks comprehensive. But these metrics describe coverage in space and frequency without addressing whether the analysis extracts temporal information from successive surveys.

Most CV anomaly detection models are trained to classify individual images: "this image shows a crack" or "this image shows normal rock face." They operate on each survey independently, without reference to what the same location looked like last week or last month. The result is a series of disconnected snapshots rather than a time-series of progressive change. A crack that is growing 2mm per week may appear "normal" in every individual snapshot while following a trajectory toward failure.

Legacy geotechnical monitoring programs solved this through human memory and continuity: the same engineer walking the same wall sections noted changes from visit to visit. When drone programs replace human continuity with CV automation, the temporal dimension is often lost unless explicitly engineered into the analysis pipeline.

The gap is the difference between "what does this look like right now?" and "how has this changed since we last looked?" Without temporal comparison, drone programs detect states but miss trajectories.

Decision Supported:

Are drone surveys detecting progressive failures, or just snapshot anomalies, and is the operation missing slow-developing hazards between surveys?

Formula

Plain Language

Count the survey areas where the CV analysis explicitly compared the current imagery to the previous survey's imagery at the same location and generated change-detection outputs. Divide by the total number of survey areas in the inspection program. Multiply by 100 to express as a percentage.

Notation

TCDR = (Survey_areas_with_temporal_comparison / Total_survey_areas) x 100

Variables

Variable Description Source
Survey_areas_with_temporal_comparison Count of discrete survey areas where the CV analysis pipeline performed explicit comparison between current and prior imagery at the same location, producing change-detection outputs (difference maps, displacement vectors, change classification) Inspection CV platform analysis logs
Total_survey_areas Count of all discrete survey areas included in the inspection program for the current cycle Drone platform flight plan and coverage records

A survey area qualifies as "with temporal comparison" only if the analysis pipeline explicitly references prior imagery and produces change-specific outputs. Simply analyzing the same area twice with snapshot models does not qualify; the comparison must be explicit and automated within the analysis workflow.

The result is a percentage between 0% and 100%. A value of 100% means every survey area benefits from temporal comparison. A value of 0% means the entire program operates in snapshot mode with no automated change detection.

Proposed formula, pending validation pilots.

At a Glance

Unit
Percentage
Trend Direction
Higher
Reporting Frequency
Monthly
Category
Inspection Fidelity & Anomaly Detection Quality
← Inspection Failure Mode Coverage Ratio (FMCR-M)Cross-Component Cascade Attribution Rate (XCAR) →

Assess Your Organization

The AI-GPS Assessment evaluates your AI governance maturity and recommends KPIs like this one for your implementation roadmap.

Take the Free AI-GPS Assessment