Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training scoring models. One method includes storing data identifying a plurality of positive and a plurality of negative training images for a query. The method further includes selecting a first image from either the positive group of images or the negative group of images, and applying a scoring model to the first image. The method further includes selecting a plurality of candidate images from the other group of images, applying the scoring model to each of the candidate images, and then selecting a second image from the candidate images according to scores for the images. The method further includes determining that the scores for the first image and the second image fail to satisfy a criterion, updating the scoring model, and storing the updated scoring model.

Abstract: A method and system of identity masking to obscure identities corresponding to face regions in an image is disclosed. A face detector is applied to detect a set of possible face regions in the image. Then an identity masker is used to process the detected face regions by identity masking techniques in order to obscure identities corresponding to the regions. For example, a detected face region can be blurred as if it is in motion by a motion blur algorithm, such that the blurred region can not be recognized as the original identity. Or the detected face region can be replaced by a substitute facial image by a face replacement algorithm to obscure the corresponding identity.

Abstract: A method and system generates and compares fingerprints for videos in a video library. The video fingerprints provide a compact representation of the spatial and sequential characteristics of the video that can be used to quickly and efficiently identify video content. Because the fingerprints are based on spatial and sequential characteristics rather than exact bit sequences, visual content of videos can be effectively compared even when there are small differences between the videos in compression factors, source resolutions, start and stop times, frame rates, and so on. Comparison of video fingerprints can be used, for example, to search for and remove copyright protected videos from a video library. Further, duplicate videos can be detected and discarded in order to preserve storage space.

Abstract: A set of reference videos is indexed to a reference index in order to facilitate matching of video content. An indexing module receives a set of reference fingerprints representing a set of reference videos and identifies keys contained in the reference fingerprints. Reference identifiers for the reference videos are stored in bins of the reference index associated with the identified keys. The bins in the reference index are sub-sampled to limit the number of reference identifiers stored in a given bin.

Abstract: A method and an apparatus localize an object part location in a digital image. The method according to one embodiment accesses digital image data containing an object part; obtains an initial position estimate for object part location of the object part; extracts a sub-image window around the initial position estimate of the object part location; calculates feature values based on pixel values within the sub-image window; and determines an updated position estimate for the object part location based on the calculated feature values.

Abstract: A sequence of images depicts a foreground object in motion. A base image is selected, and the other images in the sequence are co-registered with the base image in order to align the images to a common coordinate system. A background image and a binary foreground mask are generated from the sequence of aligned images. By applying the foreground mask to a chosen one of the aligned images, a representation of the moving object is extracted. After blurring the background image, the extracted representation may be superimposed onto the blurred background image to produce a new image.

Abstract: A method and an apparatus estimate an object part location in a digital image using feature value analysis. The method according to one embodiment accesses digital image data representing a region including an object part of a digital image; accesses reference data including class data of classes relating to predetermined positions of the object part in predetermined regions, and features that discriminate among the classes; calculates feature values for the features in the region using pixel values within the region; and determines a location estimate of the object part using the feature values and the reference data.

Abstract: A method and system of identity masking to obscure identities corresponding to face regions in an image is disclosed. A face detector is applied to detect a set of possible face regions in the image. Then an identity masker is used to process the detected face regions by identity masking techniques in order to obscure identities corresponding to the regions. For example, a detected face region can be blurred as if it is in motion by a motion blur algorithm, such that the blurred region can not be recognized as the original identity. Or the detected face region can be replaced by a substitute facial image by a face replacement algorithm to obscure the corresponding identity.

Abstract: A method and an apparatus automatically recognize or verify objects in a digital image using probability models. According to a first aspect, a method and apparatus automatically recognize or verify objects in a digital image by: accessing digital image data including an object of interest therein; detecting an object of interest in the image; normalizing the object to generate a normalized object representation; extracting a plurality of features from the normalized object representation; and applying each feature to a previously-determined additive probability model to determine the likelihood that the object of interest belongs to an existing class. In one embodiment, the previously-determined additive probability model is an Additive Gaussian Model.

Abstract: A method and an apparatus automatically detect red-eye regions in a photographic image. The method according to one embodiment accesses digital image data representing a color photographic image; detects red-eye candidate regions in the photographic image by processing the digital image data, the step of detecting red-eye candidate regions including calculating feature values for multiple positions of the photographic image, and identifying red-eye candidate regions based on the calculated feature values resulting from the calculating step; and performs face detection to determine whether an identified red-eye candidate region is associated with a face.

Abstract: A method and an apparatus localize an object part location in a digital image. The method according to one embodiment accesses digital image data containing an object part; obtains an initial position estimate for object part location of the object part; extracts a sub-image window around the initial position estimate of the object part location; calculates feature values based on pixel values within the sub-image window; and determines an updated position estimate for the object part location based on the calculated feature values.

Abstract: A method and an apparatus estimate an object part location in a digital image using feature value analysis. The method according to one embodiment accesses digital image data representing a region including an object part of a digital image; accesses reference data including class data of classes relating to predetermined positions of the object part in predetermined regions, and features that discriminate among the classes; calculates feature values for the features in the region using pixel values within the region; and determines a location estimate of the object part using the feature values and the reference data.

Abstract: A sequence of images depicts a foreground object in motion. A base image is selected, and the other images in the sequence are co-registered with the base image in order to align the images to a common coordinate system. A background image and a binary foreground mask are generated from the sequence of aligned images. By applying the foreground mask to a chosen one of the aligned images, a representation of the moving object is extracted. After blurring the background image, the extracted representation may be superimposed onto the blurred background image to produce a new image.

Abstract: A method and an apparatus automatically recognize or verify objects in a digital image using probability models. According to a first aspect, a method and apparatus automatically recognize or verify objects in a digital image by: accessing digital image data including an object of interest therein; detecting an object of interest in the image; normalizing the object to generate a normalized object representation; extracting a plurality of features from the normalized object representation; and applying each feature to a previously-determined additive probability model to determine the likelihood that the object of interest belongs to an existing class. In one embodiment, the previously-determined additive probability model is an Additive Gaussian Model.

Abstract: A method and an apparatus automatically detect red-eye regions in a photographic image. The method according to one embodiment accesses digital image data representing a color photographic image; detects red-eye candidate regions in the photographic image by processing the digital image data, the step of detecting red-eye candidate regions including calculating feature values for multiple positions of the photographic image, and identifying red-eye candidate regions based on the calculated feature values resulting from the calculating step; and performs face detection to determine whether an identified red-eye candidate region is associated with a face.