Method, apparatus and computer program product for tracking face portion

1. A method comprising:

using a processor to perform the steps of

detecting a face portion in a frame of a plurality of frames of a multimedia content, wherein detecting the face portion in the frame comprises:

providing a detection window configured to overlay onto a set of pixels in the frame;

evaluating the set of pixels bounded by the detection window based on at least one layer of classification functions for detecting the face portion in the set of pixels, wherein evaluating the set of pixels bounded by the detection window comprises: evaluating the set of pixels based on a first layer of classification function of the at least one layer of classification functions, the evaluation configured to be performed for one or more angles of orientation and further configured to compute a classification score for at least one of the one or more angles of orientation; selecting an angle with a maximum score of computed classification scores as an angle of orientation of the face portion; and detecting the face portion in the set of pixels by performing evaluation of the set of pixels based on the at least one layer of classification function, the evaluation configured to be performed for the angle of orientation of the face portion;

traversing the detection window by a first step size onto a neighboring set of pixels of the set of pixels; and

evaluating the neighboring set of pixels based on the at least one layer of classification functions for detecting the face portion in the neighboring set of pixels;

tracking the face portion in at least one subsequent frame of the frame; and

performing a color-tracking of the face portion on losing a track of the face portion in the at least one subsequent frame, where the color-tracking is performed for re-tracking the face portion in the at least one subsequent frame.

2. The method of claim 1 further comprising shifting a starting point of the detection window by a pre-defined size in successive frames of the plurality of frames for detecting the face portion.

3. The method of claim 1 , wherein the detection of the face portion is performed in the frame for one or more scales of a plurality of scales and a number of scales of the plurality of scales over which the detection of the face portion is performed is divided across the plurality of frames, the plurality of scales representing a plurality of sizes of the face portion.

4. The method of claim 1 , wherein the color-tracking comprises:

computing a horizontal luminance projection, a vertical luminance projection and a color histogram for test detections window of a plurality of test detection windows in a region of interest in a subsequent frame of the at least one subsequent frame, the region of interest corresponding to a region in the frame in which the face portion is detected;

computing a luminance projection distance for the test detection windows based on the horizontal luminance projection and the vertical luminance projection for the test detection windows and a reference horizontal luminance projection and a reference vertical luminance projection;

computing a color histogram distance for the test detection window based on the color histogram for the test detection windows and a reference color histogram; and

determining a presence of the face portion in the region of interest based on the luminance projection distance and the color histogram distance for the test detection windows.

5. The method of claim 4 , wherein the luminance projection distance is

computed for the detection windows based on a difference between the reference horizontal luminance projection and the computed horizontal luminance projection for the test detection windows, and a difference between the reference vertical luminance projection and the computed vertical luminance projection for the test detection windows.

6. An apparatus comprising:

at least one processor; and

at least one memory comprising computer program code, the at least one memory and

the computer program code configured to, with the at least one processor, cause the apparatus

at least to:

detect a face portion in a frame of a plurality of frames of a multimedia content;

track the face portion in at least one subsequent frame of the frame; and

perform a color-tracking of the face portion on losing a track of the face portion in the at least one subsequent frame, wherein the color-tracking is performed for re-tracking the face portion in the at least one subsequent frame;

wherein the apparatus is further caused, at least in part, to detect the face portion in the frame by:

providing a detection window configured to overlay onto a set of pixels in the frame;

evaluating the set of pixels bounded by the detection window based on at least one layer of classification function for detecting the face portion in the set of pixels;

traversing the detection window by a first step size onto a neighboring set of pixels of the set of pixels; and

evaluating the neighboring set of pixels based on the at least one layer of classification function for detecting the face portion in the neighboring set of pixels, and

wherein the apparatus is further caused, at least in part, to evaluate the set of pixels bounded by the detection window by:

evaluating the set of pixels based on a first layer of classification function of the at least one layer of classification function, the evaluation configured to be performed for one or more angles of orientation and further configured to compute a classification score for at least one of the one or more angles of orientation;

selecting an angle with a maximum score of computed classification scores as an angle of orientation of the face portion; and

detecting the face portion in the set of pixels by performing evaluation of the set of pixels based on the at least one layer of classification function, the evaluation configured to be performed for the angle of orientation of the face portion.

7. The apparatus of claim 6 , wherein the apparatus is further caused, at least in part, to shift a starting point of the detection window by a pre-defined size in successive frames of the plurality of frames for detecting the face portion.

8. The apparatus of claim 6 , wherein the apparatus is further caused, at least in part, to configure detecting of the face portion to be performed in the frame for one or more scales of a plurality of scales and a number of scales of the plurality of scales over which the detection of the face portion is performed is divided across the plurality of frames, the plurality of scales representing a plurality of sizes of the face portion.

9. The apparatus of claim 6 , wherein the apparatus is further caused, at least in part, to perform the color-tracking by:

computing a horizontal luminance projection, a vertical luminance projection and a color histogram for test detection windows of a plurality of test detection windows in a region of interest in a subsequent frame of the at least one subsequent frame, the region of interest corresponding to a region in the frame in which the face portion is detected;

computing a luminance projection distance for the test detection windows based on the horizontal luminance projection and the vertical luminance projection for the test detection windows and a reference horizontal luminance projection and a reference vertical luminance projection;

computing a color histogram distance based on the color histogram for the test detection windows and a reference color histogram; and

determining a presence of the face portion in the region of interest based on the luminance projection distance and the color histogram distance for the test detection windows.

10. The apparatus of claim 9 , wherein the apparatus is further caused, at least

in part, to compute the luminance projection distance for the test detection windows based on a difference between the reference horizontal luminance projection and the computed horizontal luminance projection for the test detection windows, and a difference between the reference vertical luminance projection and the computed vertical luminance projection for the test detection windows.

11. A computer program product comprising a non-transitory computer-readable storage medium, the computer-readable storage medium comprising a set of instructions, which, when executed by one or more processors, cause an apparatus to at least perform:

detect a face portion in a frame of a plurality of frames of a multimedia content;

track the face portion in at least one subsequent frame of the frame; and

perform a color-tracking of the face portion on losing a track of the face portion in the at least one subsequent frame, wherein the color-tracking is performed for re-tracking the face portion in the at least one subsequent frame, wherein the apparatus is further caused, at least in part, to detect the face portion in the frame by:

providing a detection window configured to overlay onto a set of pixels in the frame;

evaluating the set of pixels bounded by the detection window based on at least one layer of classification functions for detecting the face portion in the set of pixels;

traversing the detection window by a first step size onto a neighboring set of pixels of the set of pixels; and

evaluating the neighboring set of pixels based on the at least one layer of classification function for detecting the face portion in the neighboring set of pixels, and

wherein the apparatus is further caused, at least in part, to evaluate the set of pixels bounded by the detection window based on one or more layers of classification functions by:

evaluating the set of pixels based on a first layer of classification function of the at least one layer of classification function, the evaluation configured to be performed for one or more angles of orientation and further configured to compute a classification score for at least one of the one or more angles of orientation;

selecting an angle with a maximum score of computed classification scores as an angle of orientation of the face portion; and

detecting the face portion in the set of pixels by performing evaluation of the set of pixels based on the at least one layer of classification function, the evaluation configured to be performed for the angle of orientation of the face portion.

12. The computer program product of claim 11 , wherein the apparatus is further caused, at least in part, to shift a starting point of the detection window by a pre-defined size in successive frames of the plurality of frames for detecting the face portion.

13. The computer program product of claim 11 , wherein the apparatus is further caused, at least in part, to configure detecting of the face portion to be performed in the frame for one or more scales of a plurality of scales and a number of scales of the plurality of scales over which the detection of the face portion is performed is divided across the plurality of frames, the plurality of scales representing a plurality of sizes of the face portion.

14. The computer program product of claim 11 , wherein the apparatus is further caused, at least in part to perform the color-tracking by:

computing a horizontal luminance projection, a vertical luminance projection and a color histogram for test detection windows of a plurality of test detection windows in a region of interest in a subsequent frame of the at least one subsequent frame, the region of interest corresponding to a region in the frame in which the face portion is detected;

computing a luminance projection distance for the test detection windows based on the horizontal luminance projection and the vertical luminance projection for the test detection windows and a reference horizontal luminance projection and a reference vertical luminance projection;

computing a color histogram distance for the detection windows based on the color histogram for the test detection window and a reference color histogram; and

determining a presence of the face portion in the region of interest based on the luminance projection distance and the color histogram distance for the test detection windows.