A CNN trained on a new 14,982-image LPBF dataset reportedly classifies four defect types with ~99% accuracy, but detection/segmentation claims are not actually evaluated.
Eng Fract Mech 268:108467
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cond-mat.mes-hall 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
LabelImg: CNN-Based Surface Defect Detection
A CNN trained on a new 14,982-image LPBF dataset reportedly classifies four defect types with ~99% accuracy, but detection/segmentation claims are not actually evaluated.