Schedule Demo
Back to Technology

Accuracy & Performance Methodology

Transparency in how we measure and report platform performance. All claims are based on documented methodologies with stated limitations.

Report Writing Time Reduction

Up to 72% faster

Baseline

Manual report writing using Word/Excel templates

Measurement

Time from field data entry to finalized report

Sample Size

127 reports across 8 engineering firms

Test Conditions

  • Standard inspection reports (5-15 pages)
  • Trained users with 2+ weeks platform experience
  • Reports requiring 10-50 photos

Important Limitations

  • •Results vary based on report complexity
  • •Does not include review/approval time
  • •First-time users typically see 30-40% improvement

Defect Detection Accuracy

95%+ (mAP@0.5)

Baseline

Mean Average Precision at IoU threshold 0.5

Measurement

Standard COCO evaluation metrics

Sample Size

10,000+ labeled images in test set

Test Conditions

  • Surface-visible defects only (cracks, corrosion, spalling)
  • Images captured in adequate lighting conditions
  • Minimum 1080p resolution

Accuracy by Defect Type

Surface Cracks

96.2%

Corrosion/Rust

94.8%

Spalling/Delamination

93.1%

Paint/Coating Damage

95.5%

Structural Deformation

89.4%

Important Limitations

  • •Does not detect internal/subsurface defects
  • •Performance degrades in poor lighting or image blur
  • •Accuracy varies by defect type (see breakdown below)
  • •Not a replacement for engineering judgment

Efficiency Gain (ROI)

Up to 10x reported

Baseline

Customer-reported productivity improvements

Measurement

Self-reported survey data from pilot customers

Sample Size

12 pilot customers over 6-month period

Test Conditions

  • Teams with 5+ inspectors
  • High-volume inspection workflows (50+ inspections/month)
  • Full platform adoption (all products)

Important Limitations

  • •Self-reported data, not independently verified
  • •Results vary significantly by use case
  • •Includes time savings, not direct ROI calculation
  • •Does not account for implementation costs

Our Commitment to Transparency

We believe enterprise AI adoption requires honest communication about capabilities and limitations. All performance claims are based on documented methodologies, and we continuously update this page as we gather more data.

Questions about our methodology? Contact our accuracy team

Back to Technology