Abstract:To improve the efficiency of phenotypic data acquisition for Pareuchiloglanis anteanalis and reduce errors caused by manual measurements, an automated phenotype measurement and body-weight prediction method based on keypoint detection was developed. A total of 951 images from the ventral, lateral, and dorsal views were collected and keypoints were annotated in the COCO format. Phenotypic measurements, including traditional morphometric traits (e.g., body length, body depth, and interorbital dist