Abstract
We introduce a novel method to address intra-class imbal-
ance in 3D point cloud segmentation of wheat, focusing on
distinguishing between ear and non-ear parts. Variability in
plant structure, influenced by factors such as curvature and
shape, often leads to data imbalance which complicates seg-
mentation tasks. Our approach utilizes Monte Carlo Dropout
to identify and prioritize uncertain samples at the end of
each training epoch, employing uncertainty-driven sampling
to select samples with the lowest confidence. These samples
undergo augmentation through scaling and leaf crossover tech-
niques, enhancing their representation in the training set. Our
comparative evaluations demonstrate that this strategy signifi-
cantly improves the mean Intersection over Union (mIoU) and
segmentation accuracy, thereby increasing model robustness
for complex 3D plant structures.
ance in 3D point cloud segmentation of wheat, focusing on
distinguishing between ear and non-ear parts. Variability in
plant structure, influenced by factors such as curvature and
shape, often leads to data imbalance which complicates seg-
mentation tasks. Our approach utilizes Monte Carlo Dropout
to identify and prioritize uncertain samples at the end of
each training epoch, employing uncertainty-driven sampling
to select samples with the lowest confidence. These samples
undergo augmentation through scaling and leaf crossover tech-
niques, enhancing their representation in the training set. Our
comparative evaluations demonstrate that this strategy signifi-
cantly improves the mean Intersection over Union (mIoU) and
segmentation accuracy, thereby increasing model robustness
for complex 3D plant structures.
| Original language | English |
|---|---|
| Title of host publication | Proceedings of the 6th IEEE International Conference on Image Processing Applications and Systems |
| Publisher | IEEE Explore |
| Publication status | Accepted/In press - 12 Nov 2024 |
Keywords
- Imbalance,
- Uncertainty
- Part-segmentation
- Wheat