Research Article
Face Detection Algorithm Based on Double-Channel CNN with Occlusion Perceptron
Table 2
The average accuracy of face detection in MAFA dataset.
| Attribute | Literature [14] | Literature [16] | Literature [30] | Literature [31] | Proposed |
| Left | 3.02 | 11.8 | 17.4 | 14.2 | 19.9 | Left-front | 28.7 | 35.3 | 46.8 | 39.5 | 65 | Front | 66.8 | 70.4 | 76.8 | 72.6 | 82.2 | Right-front | 20.7 | 25.4 | 32.6 | 28.9 | 59.3 | Right | 2.07 | 6.78 | 17.5 | 9.94 | 19.3 | Weak | 59.6 | 64 | 74.9 | 67.9 | 80.6 | Medium | 44.4 | 49.6 | 65.7 | 62.3 | 72.6 | Heavy | 7.23 | 20.8 | 26.3 | 22.9 | 35.8 | Simple | 64.6 | 69.7 | 74.4 | 71.5 | 81.6 | Complex | 51.2 | 54.7 | 62.3 | 57.6 | 74.9 | Body | 25.5 | 30.3 | 43.2 | 39.6 | 65.4 | Hybrid | 10.8 | 16.2 | 21.8 | 18.5 | 28.7 | Accuracy | 62.9 | 67.4 | 73.5 | 69.7 | 80.2 | Speed (FPS) | 22 | 20 | 38 | 28 | 39 |
|
|