Rotation invariant neural network-based face detection

A survey on face detection in the wild: Past, present and future

“…In [158] a multiview face detector similar to the approach in [159,64] was proposed. In this work a face pose estimator using support vector regression (SVR) is first constructed and subsequently separate SVM face detectors one for each face pose are trained.…”

User oriented language model for face detection

“…Three different face detection algorithms were implemented for the selection of proper algorithms in this paper: AdaBoost based algorithm [13], Neural Network based algorithm [9], and Color based algorithm [4]. Some parts of the simple "if-then" style rules used for algorithm selection for the proof-ofconcept implementation are illustrated in table 3.…”

“…More rotated windows for a certain pose, for example, can be a simple and intuitive solution even though it requires more time for searching. Some algorithms provide rotation invariant methods as well [9].…”

“…They carry out the task by extracting certain properties (e.g., local features or holistic intensity patterns) of a set of training images acquired at a fixed pose (e.g., upright frontal pose). Additional algorithms have been proposed to learn these generic templates (e.g., eigenface and statistical distribution) or discriminant classifiers (e.g., neural networks, Fisher linear discriminant, sparse network of Winnows, decision trees, Bayes classifiers, support vector machines, and AdaBoost) [9,7,5,8,11].…”

“…Table 5 shows the conditions expressed to describe this problem. As a result of the algorithm selection, Neural Network based algorithm [9] was selected for the second example.…”