Determine whether processing attrition is geometry-dependent

Determine whether the 16.7% first-pass attrition rate of the maize-ear NeRF phenotyping pipeline is correlated with ear geometry by characterizing excluded ears against retained ears.

Background

Fifty of the 300 processed ears failed to reach trait extraction, including 35 excluded during holder-segmentation quality control. Because failures may arise from occlusion, ear placement, or unusual geometric structure, the retained sample may not represent the full geometric diversity of the input population.

The authors explicitly state that excluded ears have not been characterized relative to retained ears and therefore that a relationship between attrition and ear geometry remains unresolved. Establishing whether such a relationship exists is necessary for assessing selection bias and the generalizability of reported trait distributions.