Show Tag: face-detection

Select Other Tags

Li et al. present a purely engineering-based approach to active speaker localization. Their system uses Viola and Jones' object detection algorithm for face detection and cross-correlation for auditory speaker localization.

Humans can orient towards emotional human faces faster than towards neutral human faces.

The time it takes to elicit a visual cortical response plus the time to elicit a saccade from cortex (FEF) is longer than the time it takes for humans to orient towards faces.

Nakano et al. take this as further evidence for a sub-cortical (retinotectal) route of face detection.

Neurons in the monkey pulvinar react extremely fast to visually perceived faces (50ms).

Nakano et al. presented an image of either a butterfly or a neutral or emotional face to their participants. The stimuli were either grayscale or color-scale images, where color-scale images were isoluminant and only varied in their yellow-green color values. Since information from S-cones does not reach the superior colliculus, these faces were presumably only processed in visual cortex.

Nakano et al. found that their participants reacted to gray-scale emotional faces faster than to gray-scale neutral faces and to gray-scale faces faster than to gray-scale butterflies. Their participants reacted somewhat faster to color-scale faces than to color-scale butterflies, but this effect was much smaller than for gray-scale images. Also, the difference in reaction time to color-scale emotional faces was not significantly different from that to color-scale neutral faces.

Nakano et al. take this as further evidence of sub-cortical face detection and in particular of emotional sub-cortical face detection.

Humans can orient towards human faces faster than towards other visual stimuli (within 100ms).

Newborns track schematic, face-like visual stimuli in the periphery, up to one month of age. They start tracking such stimuli in central vision after about 2 months. and stop after 5.

According to Johnson and Morton, there are two visual pathways for face detection: the primary cortical pathway and one through SC and pulvinar.

The cortical pathway is called CONLEARN and is theorized to be plastic, whereas the sub-cortical pathway is called CONSPEC and is thought to be fixed and genetically predisposed to detect conspecific faces.

Masking visual face stimuli---ie. presenting faces visually for too short to detect them consciously, then presenting a masking stimulus---can evoke measurable changes in conductance.

Masking visual face stimuli can evoke responses in sc, pulvinar, and amygdala.

The right amygdala responded differently to masked than to unmasked stimuli, while the left did not in Morris et al.'s experiments.

Many visual person detection methods use one feature to detect people, create a histogram for the strength of that feature across the image. They then compute a likelihood for a pixel or region by assuming a Gaussian distribution of distances of pixels or histograms belonging to a face. This distribution has been validated in practise (for certain cases).

Neurons at later stages in the hierarchy of visual processing extract very complex features (like faces).

There are reports of highly selective, purely visual cells in the putamen. One report is of a cell which responded best to a human face.

Sanchez-Riera et al. use the Waldboost face detection system for visual processing.

Yan et al. use the standard Viola-Jones face detection algorithm for visual processing.

Viola and Jones presented a fast and robust object detection system based on

  1. a computationally fast way to extract features from images,
  2. the AdaBoost machine learning algorithm,
  3. cascades of weak classifiers with increasing complexities.

Newborn children prefer to look at faces and face-like visual stimuli.

The fact that visual cortex is not fully developed at birth, but newborn children prefer face-like visual stimuli to other visual stimuli could be explained by the presence of a subcortical face-detector.

The fact that visual cortex is not fully developed at birth, but newborn children prefer face-like visual stimuli to other visual stimuli could be explained by the presence of a subcortical face-detector.

In one experiment, newborns reacted to faces only if they were (exclusively) visible in their peripheral visual field, supporting the theory that the sub-cortical pathway of visual processing plays a major role in orienting towards faces in newborns.

SC has been implicated as part of a subcortical visual pathway which may drive face detection and orienting towards faces in newborns.

The subcortical visual pathway which may drive face detection and orienting towards faces in newborns hypothesized by Johnson also includes amygdala and pulvinar.

According to the hypothesis expressed by Johnson, amygdala, pulvinar, and SC together form a sub-cortical pathway which detects faces, initiates orienting movements towards faces, and activates cortical regions.

This implies that this pathway may be important for the development of the `social brain', as Johnson puts it.

Visual processing of potentially affective stimuli seems to be partially innate in primates.

Cells in the amygdala respond to faces and parts of faces. Some react exclusively to faces.

There are cells in inferotemporal cortex which respond to (specific views on / specific parts of) faces, hands, walking humans and others.

Pitti et al. claim that their model explains preference for face-like visual stimuli and that their model can help explain imitation in newborns. According to their model, the SC would develop face detection through somato-visual integration.

Pitti et al.'s claim of predicting with their model face-detectors in the developing SC and explaining preference for visual stimuli is problematic.

First, it is implausible that the somato-visual map created through multi-sensory learning would map the location at which a child sees facial features to the same neurons that respond to corresponding features in the child's own face. A child does not see a mouth where it sees a ball or hand touching its own mouth.

Secondly, their paper at least does not explain why their model should develop preference for points that correspond with their own facial features. The only reason that this could be happening I see is that their grid model of the child's face has a higher density of nodes around eyes, nose, and mouth, such that those regions are denser in the resulting map. But this would not explain why the configuration of features corresponding to a face would have higher saliency.