Sugarcane single eye-bud detection: annotated image datasets (cultivars COJN9509 and Co86032)

This deposit contains the two frozen image archives. COJN9509_frozen_2026-09-04_upload.zip : the in-domain dataset: 1,502 RGB images of sugarcane cultivar COJN9509 captured at three controlled camera heights (10/12/14 cm), hand-annotated in Ultralytics YOLO format with three classes (EYE_BUD, NODE, INTERNODE), split train (1,053) / valid (224) / test (225). SHA-256: 06D7A7108EE236D83B15E13A727B22027D046C42339D1D691610DA5165C0EC73 ; data.yaml CRC-32: D6C6A887. Co86032_frozen_2026-09-04_upload.zip: the external cross-cultivar test set: 414 images of cultivar Co86032 at 30 cm, used only for the no-retraining transfer evaluation. SHA-256: 3DB95E6C5358E75A8D37A774D932CE2F76089F89E3DFA8BED427DC4E7668FE59 ; data.yaml CRC-32: E4348705. Every measurement in publication reads these snapshots and verifies the fingerprint above before use. Trained detector weights are not redistributed (the specialist models were trained with Ultralytics, AGPL-3.0); the analysis, training, and verification code, and the results the publication reports, are available in the GitHub repository cited in the article. Please cite this deposit together with the associated article.