.png){kind=icon}
.png){kind=icon}
Return
Classification of Copper-Nickel ore based on Microwave-Induced thermal Excitation and infrared image Generation: Model generalization analysis
.png){kind=icon}
PRE
AI
K
K
C
C
P
Z
G
C
DOI：10.1016/j.mineng.2026.110727.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• Generated infrared images supported stable in-domain classifier training. • Training using only generated data showed limited generalization to real infrared images. • Mixed training with a small number of real samples significantly improved real-scene classification. • Moderate generated-to-real ratios were more conducive to stable optimization. • A small number of real samples anchored the target distribution and reduced domain shift.
Keywords:
Microwave induction
Infrared imaging
Generative adversarial networks
Copper-nickel ore-waste rock classification
Mixed training
Model generalization
Journal
IF:
5
Papers:
8.1K
Citations:
2.6W
Organization
Cited Papers