Video content analysis for automatic demographics recognition of users and videos

Abstract

A video demographics analysis system selects a training set of videos to use to correlate viewer demographics and video content data. The video demographics analysis system extracts demographic data from viewer profiles related to videos in the training set and creates a set of demographic distributions, and also extracts video data from videos in the training set. The video demographics analysis system correlates the viewer demographics with the video data of videos viewed by that viewer. Using the prediction model produced by the machine learning process, a new video about which there is no a priori knowledge can be associated with a predicted demographic distribution specifying probabilities of the video appealing to different types of people within a given demographic category, such as people of different ages within an age demographic category.

1 . A computer-implemented method of generating a prediction model for videos, comprising:
 receiving a plurality of videos from a video repository, each video having an associated list of viewers;   for each video, creating a demographic distribution for at least one demographic attribute based at least in part on viewer demographic data associated with viewers of the video;   for each video, generating feature vectors based at least in part on the content of the video;   generating a prediction model that correlates the feature vectors for the videos and the demographic distributions; and   storing the prediction model.   
     
     
         2 . The computer-implemented method of  claim 1 , wherein the demographic attribute is one of age and gender. 
     
     
         3 . The computer-implemented method of  claim 1 , wherein the demographic attribute is one of occupation, household income, and location. 
     
     
         4 . The computer-implemented method of  claim 1 , wherein the prediction model is generated using support vector machines. 
     
     
         5 . The computer-implemented method of  claim 1 , further comprising altering the generated feature vectors using a dimensionality reduction algorithm. 
     
     
         6 . The computer-implemented method of  claim 1 , wherein the generated feature vectors include features vectors generated based on audio content of the video and vectors generated based on visual content of the video. 
     
     
         7 . The computer-implemented method of  claim 1 , wherein the feature vectors are generated at least in part on metadata associated with the video. 
     
     
         8 . The computer-implemented method of  claim 1 , further comprising:
 performing object segmentation on a frame of the video, thereby identifying a visual object of the frame;   wherein generating feature vectors based at least in part on the content of the video comprises generating feature vectors for the identified visual object.   
     
     
         9 . A computer-implemented method for determining demographics of a video, comprising:
 storing a prediction model that correlates viewer demographic attributes with feature vectors extracted from videos viewed by viewers, wherein the viewer demographic attributes include age and gender;   receiving a video;   generating from content of the video a set of feature vectors; and   identifying demographic attribute values by applying the prediction model to the generated set of feature vectors.   
     
     
         10 . The computer-implemented method of  claim 8 , wherein identifying demographic attribute values comprises:
 identifying a set of feature vectors of the prediction model that is most similar to the generated set of feature vectors; and   identifying, in the prediction model, demographic attribute values most strongly correlated with the identified feature vectors.   
     
     
         11 . A computer-implemented method for identifying demographics associated with a viewer, comprising:
 storing a prediction model that correlates viewer demographic attributes with feature vectors generated from videos viewed by viewers;   identifying a set of videos viewed by a given viewer;   generating, from content of the set of videos, feature vectors;   applying the feature vectors to the prediction model to identify viewer demographic attribute values most strongly correlated with the feature vectors of the prediction model; and   identifying viewer demographic attribute values most strongly correlated with the given viewer based at least in part on the identified viewer demographic attribute values.   
     
     
         12 . A computer-implemented method for identifying videos associated with given demographic attribute values, comprising:
 storing a prediction model that correlates viewer demographic attributes with feature vectors generated from videos viewed by viewers;   receiving a plurality of videos;   for each video of the plurality of videos:
 generating feature vectors from the video; 
 applying the feature vectors generated from the video to the prediction model to identify viewer demographic attribute values most strongly correlated with the feature vectors of the prediction model; 
 storing the identified viewer demographic attribute values in association with the video; 
   selecting videos having highest values for the given demographic attribute values; and   displaying identifiers of the selected videos.   
     
     
         13 . A computer readable storage medium storing a computer program executable by a processor for generating a prediction model for videos, the actions of the computer program comprising:
 receiving a plurality of videos from a video repository, each video having an associated list of viewers;   for each video, creating a demographic distribution for at least one demographic attribute based at least in part on viewer demographic data associated with viewers of the video;   for each video, generating feature vectors based at least in part on the content of the video;   generating a prediction model that correlates the feature vectors for the videos and the demographic distributions; and   storing the prediction model.   
     
     
         14 . The computer readable storage medium of  claim 12 , wherein the generated feature vectors include features vectors generated based on audio content of the video and vectors generated based on visual content of the video. 
     
     
         15 . The computer readable storage medium of  claim 12 , wherein the prediction model is generated using support vector machines 
     
     
         16 . A computer system for generating a prediction model for videos, comprising:
 a video repository storing a plurality of videos, each video having an associated list of viewers;   a video analysis server adapted to:
 receive a plurality of videos from the video repository; 
 for each video, create a demographic distribution for at least one demographic attribute based at least in part on viewer demographic data associated with viewers of the video; 
 for each video, generate feature vectors based at least in part on the content of the video; 
 generate a prediction model that correlates the feature vectors for the videos and the demographic distributions; and 
 store the prediction model. 
   
     
     
         17 . The computer system of  claim 16 , wherein the demographic attribute is one of age and gender. 
     
     
         18 . The computer system of  claim 16 , wherein the prediction model is generated using support vector machines. 
     
     
         19 . The computer system of  claim 16 , wherein the generated feature vectors include features vectors generated based on audio content of the video and vectors generated based on visual content of the video. 
     
     
         20 . The computer system of  claim 16 , wherein the feature vectors are generated at least in part on metadata associated with the video.