Answers are grounded in this paper and cite their sources.
Aluminium-lithium alloys support lightweight aerospace structures, but pitting creates localized material loss and potential fatigue-crack initiation sites. Quantifying this damage requires individual pit boundaries linked to physical depth. Here we develop PitRPM to connect instance recognition with geometric measurement in confocal height fields of 2195-T8 alloy. Height, local depth residual and gradient channels feed a mask segmenter and an optional vision-language proposal branch. Accepted masks are measured on the original height field. Across 120 fields containing 1,138 reviewed pits, field-level validation yielded an F1 score of 0.826 and mean mask intersection-over-union of 0.758. Matched-pit relative errors were 31.8% for projected area, 0.8% for maximum depth and 5.2% for mean depth. Additional proposals increased recall with a corresponding reduction in precision. The framework connects automated pit recognition to physical descriptors and distinguishes the boundary sensitivity of area from the pit-bottom sensitivity of depth.
No takes yet. Share an insight, caveat, or question.
Liu et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: