Table 2 Comparison of Acc metrics between the proposed method and state-of-the-art approaches for mural figures on the test set
From: A Mamba based vision transformer for fine grained image segmentation of mural figures
Acc | |||||||
|---|---|---|---|---|---|---|---|
Method | Other | Face | Headwear | Bracelet | Armlet | mAcc | aAcc |
PSPNet | 99.81 | 94.54 | 75.13 | 40.08 | 38.38 | 69.59 | 99.27 |
OCRNet | 99.80 | 91.82 | 71.80 | 38.76 | 43.22 | 69.08 | 99.20 |
Segmenter | 99.70 | 90.10 | 56.79 | 21.11 | 40.47 | 61.64 | 98.90 |
SegFormer | 99.79 | 95.43 | 74.70 | 48.23 | 45.48 | 72.73 | 99.29 |
DDRNet | 99.87 | 91.53 | 61.39 | 34.99 | 31.28 | 63.81 | 99.13 |
SAN | 99.66 | 95.04 | 69.25 | 51.58 | 27.07 | 68.52 | 99.07 |
Ours | 99.83 | 94.01 | 72.37 | 47.32 | 59.44 | 74.59 | 99.31 |
