Class-Balanced Random Patch Training to Address Class Imbalance in Tiling-Based Farmland Classification
Satellite image-based farmland classification plays an essential role in agricultural monitoring. However, typical tiling-based classification approaches, which extract patches at fixed offsets within each image during training, often suffer from structural issues such as patch duplication, limiting...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-06-01
|
| Series: | Applied Sciences |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2076-3417/15/13/7056 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|