To search, Click below search items. |
| All Published Papers Search Service |
|
|
|
Title |
Effective Detection of Human Face with GentleBoost Approach |
|
Author |
Lam Thanh Hien, Do Nang Toan, Ha Manh Toan, Thanh-Lam Nguyen |
| Citation |
Vol. 16 No. 12 pp. 103-108 |
|
Abstract |
Detecting human face has been a popular but complex research topic which has successfully attracted the special attention of academic scholars and experts due to its widespread applications in practice. However, several existing methods fail to work in real time or with low detection rate, or large-slanting angles of faces, etc. To overcome such drawback, this study proposes an innovative approach with GentleBoost to effectively detect human faces in images. Our proposed algorithm with two consecutive phases namely “learning phase” and “detecting phase” has shown its effectiveness. Specifically, its performance has been well validated through hundreds of images collected from reliable databases and self-recorded sources. It is found that though the detection rate from our approach is lower than that of traditional Haar-AdaBoost, ours still provides satisfactory results in terms of precision and recall. More importantly, it is about 6 times faster than that of traditional Haar-AdaBoost, promising a great potential to be integrated into practical applications that need to detect human faces in real time. |
|
Keywords |
Detect human face, GentleBoost, Haar-AdaBoost, Real time detection, Precision, Recall |
|
URL |
