Research Article
DetReco: Object-Text Detection and Recognition Based on Deep Neural Network
Table 2
Performance of different detection frameworks on different testing datasets.
| Detection frameworks | VOC2007 | SCUT_FORU | ICDAR2013 | mAP | Car | Bus | Person | Text | Text |
| Fast R-CNN [19] | 70.0 | 78.6 | 81.6 | 69.9 | — | — | Faster R-CNN VGG-16 [20] | 73.2 | 84.7 | 83.1 | 76.7 | — | — | SSD300 [16] | 74.3 | 84.2 | 83.0 | 76.2 | — | — | SSD512 [16] | 76.8 | 87.5 | 86.2 | 79.7 | — | — | YOLOv3 416 × 416 | 77.9 | 90.4 | 85.4 | 85.7 | 70.0 | 64.2 | YOLOv3 544 × 544 | 78.3 | 91.5 | 89.4 | 86.8 | 70.0 | 70.0 | YOLOv3 608 × 608 | 77.9 | 90.4 | 85.4 | 85.7 | 70.0 | 64.2 |
|
|