Table 1 Experimental results
From: Improved Viola-Jones face detection algorithm based on HoloLens
Face image database | Algorithm mentioned in this paper | Microsoft Azure Face API | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
Before acceleration and optimization | After acceleration and optimization | |||||||||
Name | Feature | Size | Qty. | Missing (Qty.) | Time (s) | Missing (Qty.) | Time (s) | Missing (Qty.) | ime (s) | Total time (s) |
Orl | Expression change | 92 × 112 | 400 | 21 | 37 | 21 | 10.6 | 111 | 67.5 | 1467.5 |
Yale | Illumination expression glasses | 60 × 60 | 165 | 7 | 5 | 7 | 1.3 | 44 | 26.5 | 604 |
Ar | Illumination expression scarf | 72 × 99 | 2600 | 352 | 173 | 352 | 43 | 631 | 425.3 | 9526.3 |
Stanford | Face calibration | 139 × 139 | 400 | 3 | 102.9 | 3 | 20.1 | 129 | 69.6 | 1469.6 |
Jaffe | Expression of Asian woman | 153 × 153 | 213 | 0 | 64.4 | 0 | 17.6 | 0 | 47 | 792.4 |
cit | Color illumination change | 179 × 118 | 450 | 14 | 98.3 | 14 | 24.9 | 16 | 96.7 | 1671.9 |