Predicted measurements did not differ significantly from ground truth.

Method / resourceobservedoriginal resultno effectnot significant (as stated)measured

Claim · model-written summary

Taxon: mandarint0
Outcome: YOLOv8-predicted measurementse0
Comparator: ground truth datae0

Not reportedMaterialGeneCompoundOrgan / stageInteractorInteractor roleSecond interactorSecond interactor roleTreatmentSettingDirectionValueQualifier

Verbatim

e0
“nonsignificant differences between YOLOv8-predicted measurements and ground truth data”
✓ located verbatim in the source · abstract, characters 1221–1307

t0
“mandarin”
✓ located verbatim in the source · title, characters 45–53

Title

Deep learning based instance segmentation of mandarin fruit slices for precision assessment and morphological quantification

Anh Tuan Le; Jinhyun Ahn; Thanh Tuan Thai Scientific Reports 2026. doi:10.1038/s41598-026-46784-4

Other claims from this paper (10)

Method / resource YOLOv8 bounding box precision was about 0.997.
Method / resource YOLOv8 bounding box recall was about 1.00.
Method / resource YOLOv8 bounding box mAP50 was about 0.995.
Method / resource YOLOv8 bounding box mAP50-95 was about 0.922.
Method / resource Semantic segmentation precision was 0.997.
Method / resource Semantic segmentation recall was about 1.00.
Method / resource Semantic segmentation mAP50 was about 0.995.
Method / resource Semantic segmentation mAP50-95 was 0.965.
Observation Training and validation losses converged steadily.
Method / resource Predicted measurements had low mean absolute error.

Record rev-9eb856d49d69fc8c77811a7ca2fb9894eb7950c16811f259bf03c71654f64c56 · schema claims-v0.9 · position 10 in the model’s answer