Research Article
Multimodal Semantics Extraction from User-Generated Videos
Table 2
Classification of events according to scene, genre, and layout.
| Event | Indoor/outdoor scene classification | Event genre classification | Layout classification | Ground truth | Proposed method | Ground truth | Proposed method | Ground truth | Proposed method |
| Football match 1 | Outdoor | Outdoor | Sport | Sport | Stadium | Stadium | Football match 2 | Outdoor | Outdoor | Sport | Sport | Stadium | Stadium | Football match 3 | Outdoor | Outdoor | Sport | Sport | Stadium | Stadium | Ice-hockey match 1 | Indoor | Outdoor | Sport | Sport | Stadium | Stadium | Ice-hockey match 2 | Indoor | Indoor | Sport | Sport | Stadium | Stadium | Concert 1 | Outdoor | Outdoor | Live music | Live music | Nonstadium | Nonstadium | Concert 2 | Outdoor | Outdoor | Live music | Live music | Nonstadium | Stadium | Concert 3 | Indoor | Indoor | Live music | Live music | Nonstadium | Nonstadium | Concert 4 | Indoor | Indoor | Live music | Live music | Nonstadium | Nonstadium |
| Classification accuracy (%) | — | 88.9 | — | 100 | — | 88.9 |
|
|