Research Article
Soccer Ball Detection by Comparing Different Feature Extraction Methodologies
Table 2
Ball recognition results using different features on the whole dataset.
| | Ball | Number | TP
| FN
| TN
| FP
| Detection | | features | of input | rate (%) |
| | Wavelet HAAR | 64 (8 8) | 1179 | 58 | 1028 | 209 | 89.20% | | Wavelet DB3 | 121 (11 11) | 1166 | 71 | 1060 | 177 | 89.97% | | Histogram | 256 | 1174 | 63 | 958 | 279 | 86.17% | | SIFT (BoW) | 20 | 621 | 616 | 1059 | 178 | 67.90% | | PCA | 30 | 999 | 238 | 890 | 347 | 76.35% |
|
|
