Aegeaeobuthus gibbosus Image Dataset for Few-Shot Sex Classification

Aegeaeobuthus gibbosus Image Dataset for Few-Shot Sex Classification This dataset contains 198 high-resolution images of 99 individual scorpions of the species Aegeaeobuthus gibbosus (Brullé, 1832), collected from multiple locations in Turkey. Each individual is photographed from two anatomical views (dorsal and ventral) under controlled laboratory conditions. The dataset is designed to support few-shot learning research on sex classification in venomous arthropods, where specimen collection is constrained by safety, ecological, and logistical factors. Dataset composition: Female: 50 individuals × 2 views = 100 images Male: 49 individuals × 2 views = 98 images Total: 99 individuals, 198 images Directory structure: female-dorsal/ — 50 dorsal-view images of female specimens female-ventral/ — 50 ventral-view images of female specimens male-dorsal/ — 49 dorsal-view images of male specimens male-ventral/ — 49 ventral-view images of male specimens female.txt and male.txt — dorsal-ventral pairing files (one individual per line, in the format DORSAL_ID-VENTRAL_ID) README.md — full documentation Important — individual-level splits: Because dorsal and ventral images of the same specimen are highly correlated, train/test splits must be performed at the individual level (not the image level) to avoid data leakage. The pairing files are provided to support correct individual-level splitting. Image acquisition: Laboratory setting, controlled lighting, white background, JPEG format. Ethical note: Aegeaeobuthus gibbosus is not a protected species under Turkish wildlife regulations; no special collection permit was required. This dataset accompanies the manuscript "Few-Shot Learning for Sex Classification in Venomous Scorpions: A Framework for Data-Scarce Biological Research" (under review). The associated code repository will be made available upon publication.