Abstract: The specification relates to providing an image space. The image space represents a first sampling of images in increasing distance from a seed image. The first sampling shows a number of images an initial distance value from the seed image and representative images of image groups a distance value that is different from the initial distance value from the seed image. The system is capable of browsing and modifying the image space responsive to at least one input. When modified, the system provides a second sampling of the images in increasing distance from an image related to a target image. The second sampling shows a number of images a certain distance value from the image related to the target image and representative images of image groups a distance value that is different from the certain distance value from the image related to the target image.

Abstract: This disclosure relates to transformation invariant media matching. A fingerprinting component can generate a transformation invariant identifier for media content by adaptively encoding the relative ordering of signal markers in media content. The signal markers can be adaptively encoded via reference point geometry, or ratio histograms. An identification component compares the identifier against a set of identifiers for known media content, and the media content can be matched or identified as a function of the comparison.

Abstract: A method and system generates and compares fingerprints for videos in a video library. The video fingerprints provide a compact representation of the temporal locations of discontinuities in the video that can be used to quickly and efficiently identify video content. Discontinuities can be, for example, shot boundaries in the video frame sequence or silent points in the audio stream. Because the fingerprints are based on structural discontinuity 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. Furthermore, duplicate videos can be detected and discarded in order to preserve storage space.

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: 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: This disclosure relates to transformation invariant media matching. A fingerprinting component can generate a transformation invariant identifier for media content by adaptively encoding the relative ordering of signal markers in media content. The signal markers can be adaptively encoded via reference point geometry, or ratio histograms. An identification component compares the identifier against a set of identifiers for known media content, and the media content can be matched or identified as a function of the comparison.

Abstract: Methods, systems, and articles of manufacture for detecting placeholder images are disclosed. These include, accessing a collection of digital images, clustering the digital images to generate at least one of a plurality of exact-duplicate image clusters and a plurality of near-duplicate image clusters, and selecting one or more placeholder image clusters from at least one of the plurality of exact-duplicate image clusters or the plurality of near-duplicate image clusters.

Abstract: A full-length fingerprint is generated for a video that represents characteristics of the entire duration of the video in a small amount of data. A plurality of subfingerprints for the video, each subfingerprint encoding features of a corresponding segment of the video, is accessed. A plurality of subhistograms is generated for the video, each subhistogram encoding a frequency of subfingerprint features for subfingerprints included in a subset of the plurality of accessed subfingerprints. A master histogram for the video is generated that encodes a frequency of subhistogram features for the plurality of generated subhistograms for the video. A hashing procedure is applied to the master histogram to produce the full-length fingerprint. Clustering can be performed on full-length video fingerprints for videos in a database to detect near-duplicate videos.

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: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for receiving a hash vector r, a vector of locality-sensitive hash values, each hash value being an element of the hash vector r, each element having an index position; and generating a compact vector v corresponding to the hash vector r, wherein the compact vector v is a vector of compact elements each having an index position, wherein each compact element corresponds to the element of the hash vector r having the same index position, and wherein each compact element is a b-bit integer selected from the set of all b-bit integers {0, 1, . . . , 2b?1} based on the corresponding hash element.

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 computer-implemented method is disclosed for generating a signature representing an input bit vector. A signature generator generates a primary min-hash value based on a primary permutation from a sequence of permutation blocks. If the primary min-hash value is lower than a threshold value, a secondary min-hash value is generated based on a secondary permutation from the same permutation block. The signature generator then determines one or more signature values based on the primary min-hash value, the secondary min-hash value or both. The one or more signature values are stored as elements of the signature.

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 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 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 robust hashing method is applied to media data (e.g., video, image, and/or audio data), producing a hash output that is robust with respect to at least one attribute of the media data. A histogram is generated for the media data and the histogram is hashed using a weighted hashing procedure. The histogram can be derived from a plurality of randomized versions of the media file, each randomized version of the media file altered to a random extent with respect to the attribute. The histogram can also be derived from a plurality of feature descriptors computed for the media data that are coarsely encoded with respect to the attribute. The weighted hashing procedure includes assigning a weight to components of the histogram and applying a plurality of hash functions to a number of versions of each component, the number of versions based on the assigned weight.

Abstract: A full-length fingerprint is generated for a video that represents characteristics of the entire duration of the video in a small amount of data. A plurality of subfingerprints for the video, each subfingerprint encoding features of a corresponding segment of the video, is accessed. A plurality of subhistograms is generated for the video, each subhistogram encoding a frequency of subfingerprint features for subfingerprints included in a subset of the plurality of accessed subfingerprints. A master histogram for the video is generated that encodes a frequency of subhistogram features for the plurality of generated subhistograms for the video. A hashing procedure is applied to the master histogram to produce the full-length fingerprint. Clustering can be performed on full-length video fingerprints for videos in a database to detect near-duplicate videos.

Abstract: Videos are received and indexed based on fingerprints generated for the videos, using lookup keys which correspond to sub portions of the reference fingerprints. Specifically, a set of lookup keys is selected such that the number of reference fingerprints such that the clumping associated with the associated each of the lookup keys is minimized. Clumping occurs when a number of reference identifiers associated with a lookup key exceeds a maximum value.

Abstract: Proprietary rights logos are detected in a video. The video is divided into a plurality of regions that are analyzed for generic proprietary rights logo features. A confidence mask is generated that comprises a plurality of scaling factors, each scaling factor corresponding to a region of the video and indicating a likelihood that the corresponding region of the video includes a proprietary rights logo. The scaling factors of the confidence mask are applied to the video data to generate an altered video. The altered video is analyzed to determine a confidence measure that the video includes a reference proprietary rights logo.

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.