Anzeige
Erschienen in:
2021 | OriginalPaper | Buchkapitel
HoG Multi-face Detection
verfasst von : Leandro Duarte, Cláriton Bernadelli
Erschienen in: Proceedings of the 5th Brazilian Technology Symposium
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
Abstract
Human face detection is important in many applications, such as human–machine interface, automatic surveillance, and facial recognition. This work exposes a solid and general face detection system capable of detecting multiple faces in the same image, even in low light situations and chaotic backgrounds. The detection system uses a representation of the Gaussian pyramid and evaluates it in all scales the existence of faces using descriptors of HoG characteristics and linear classifiers SVM. The system shows that the gradient distribution in the face contours is sufficiently discriminative to distinguish faces and non-faces and the use of cascade detectors improves overall system performance by decreasing the number of false positives. Employing experimental tests, the methodology was applied to facial and non-facial test images, allowing the evaluation of the effectiveness of the face detection system and the influence of adjustable parameters on the accuracy and performance of the system.