Skip to main content
. 2021 Dec 2;12:780180. doi: 10.3389/fpls.2021.780180

Table 2.

Comparison of the proposed convolutional neural network (CNN) relative to two other common CNN architectures.

Method Parameters Time (m) IoU Loss Acc.
U-Net ~16,482,000 200 0.78 0.18 0.98
Attention U-Net ~17,450,000 343 0.72 0.18 0.97
Proposed method ~8,114,000 176 0.84 0.16 0.98