Research Article
Mutual Positioning Method in Unknown Indoor Environment Based on Visual Image Semantics
Table 1
The recognition accuracy of R-FCN for different targets.
| Semantic target | Correct number | Wrong number | Unrecognized number | Accuracy |
| Door | 724 | 7 | 6 | 98.23% | Exhibition board | 678 | 5 | 7 | 98.26% | Poster | 375 | 3 | 4 | 98.17% | Window | 297 | 4 | 3 | 97.70% | Heating | 266 | 0 | 2 | 99.25% | Hydrant | 102 | 2 | 2 | 96.23% | Exit light | 86 | 0 | 3 | 96.63% | Vent | 47 | 2 | 0 | 95.92% | Ashbin | 39 | 0 | 1 | 97.50% |
|
|