Table 4 The performance of three different backbones on the test dataset. “R50”, “R101”, and “R101 DCN” refer to the ResNet-50, ResNet-101, and ResNet-R101 DCN, respectively. The bold fonts indicate the best results.
From: Weakly supervised learning through box annotations for pig instance segmentation
model | Param. | Box detection (%) | Mask segmentation (%) | ||||||
|---|---|---|---|---|---|---|---|---|---|
\(\textrm{AP}\) | \(\textrm{AP}_{0.5}\) | \(\textrm{AP}_{0.75}\) | \(\textrm{AP}_l\) | \(\textrm{AP}\) | \(\textrm{AP}_{0.5}\) | \(\textrm{AP}_{0.75}\) | \(\textrm{AP}_l\) | ||
R50 | 68.2M | 89.22 | 98.24 | 95.76 | 90.10 | 85.29 | 98.32 | 96.69 | 86.90 |
R101 | 102.4M | 89.78 | 98.25 | 95.73 | 90.69 | 85.51 | 98.34 | 96.79 | 87.17 |
R101 DCN | 102.4M | 89.81 | 98.30 | 95.81 | 90.64 | 85.26 | 98.36 | 96.22 | 86.87 |
