Table 2 Evaluation and comparison on the Linemod dataset in terms of the 2D projection metric. With refinement methods are marked with *.
From: MagicCubePose, A more comprehensive 6D pose estimation network
Method | BB8* | BB8 | Tekin33 | Our | |
|---|---|---|---|---|---|
Ape | 96.6 | 95.3 | 85.2 | 92.10 | 98.76 |
Benchvise | 90.1 | 80.0 | 67.9 | 95.06 | 97.48 |
Cam | 86.0 | 80.9 | 58.7 | 93.24 | 98.82 |
Can | 91.2 | 84.1 | 70.8 | 97.44 | 96.75 |
Cat | 98.8 | 97.0 | 84.2 | 97.41 | 99.20 |
Driller | 80.9 | 74.1 | 73.9 | 79.41 | 96.63 |
Duck | 92.2 | 81.2 | 73.1 | 94.65 | 98.40 |
Eggbox * | 91.0 | 87.9 | 83.1 | 90.33 | 100 |
Glue * | 92.3 | 89.0 | 74.2 | 96.53 | 94.20 |
Holepuncher | 95.3 | 90.5 | 78.9 | 92.86 | 98.29 |
Iron | 84.8 | 78.9 | 83.6 | 82.94 | 97.65 |
Lamp | 75.8 | 74.4 | 64.0 | 76.87 | 94.72 |
phone | 85.3 | 77.7 | 60.6 | 86.07 | 94.81 |
Average | 89.3 | 83.9 | 73.7 | 90.37 | 97.36 |
