Abstract: A MapReduce-based training framework exploits both data parallelism and model parallelism to scale training of complex models. Particular model architectures facilitate and benefit from use of such training framework. As one example, a machine-learned model can include a shared feature extraction portion configured to receive and process a data input to produce an intermediate feature representation and a plurality of prediction heads that are configured to receive and process the intermediate feature representation to respectively produce a plurality of predictions. For example, the data input can be a video and the plurality of predictions can be a plurality of classifications for content of the video (e.g., relative to a plurality of classes).

Abstract: A MapReduce-based training framework exploits both data parallelism and model parallelism to scale training of complex models. Particular model architectures facilitate and benefit from use of such training framework. As one example, a machine-learned model can include a shared feature extraction portion configured to receive and process a data input to produce an intermediate feature representation and a plurality of prediction heads that are configured to receive and process the intermediate feature representation to respectively produce a plurality of predictions. For example, the data input can be a video and the plurality of predictions can be a plurality of classifications for content of the video (e.g., relative to a plurality of classes).

Abstract: A computer implemented method, apparatus, and computer program product code for temporal, event-based video fingerprinting. In one embodiment, events in video content are detected. The video content comprises a plurality of video frames. An event represents discrete points of interest in the video content. A set of temporal, event-based segments are generated using the events. Each temporal, event-based segment is a segment of the video content covering a set of events. A time series signal is derived from each temporal, event-based segment using temporal tracking of content-based features of a set of frames associated with the each temporal, event-based segment. A temporal segment based fingerprint is extracted based on the time series signal for the each temporal, event-based segment to form a set of temporal segment based fingerprints associated with the video content.

Abstract: The present invention is directed to a method and apparatus for assisting in rating and filtering multimedia content, such as images, videos and sound recordings. One embodiment comprises a computer implemented method for rating the objectionability of specified digital content that comprises one or more discrete content items, wherein the method includes the step of moving the specified content to one or more filtering stages in a succession of filtering stages. After the specified content is moved to a given one of the filtering stages, a rating procedure is carried out to determine whether a rating can be applied to one or more of the content items, and if so, a selected rating is applied to each of the one or more content items. The method further comprises moving content items of the specified content to the next stage in the succession after the given stage, when at least one content item of the specified content remains without rating, after the rating procedure at the given stage.

Abstract: A method and apparatus is disclosed that receives an input narrative in a particular media format, and produces therefrom a multimedia enriched version of the input narrative. In one embodiment, a method is provided for enriching an input narrative with multimedia content, wherein the method includes the step of selectively segmenting the input narrative to form queries pertaining to different portions or elements of the input narrative. The method further comprises retrieving multimedia artifacts from one or more multimedia repositories, wherein each retrieved multimedia artifact is associated with one of the queries, and selecting a subset of the retrieved multimedia artifacts, wherein the subset includes artifacts comprising different types of multimedia content. The input narrative is then combined with respective multimedia artifacts of the subset, in order to provide an enriched and enhanced narrative.

Abstract: A computer implemented method, apparatus, and computer usable program code for multimodal cross-vocabulary mapping. A corpus of multimodal content is annotated simultaneously using annotations from a plurality of vocabularies to form a set of common annotations. Relationships between a first vocabulary associated with a first modality and a second vocabulary associated with a second modality are identified using the set of common annotations to form a multimodal vocabulary mapping. Items in the first vocabulary associated with the first modality are mapped to items in the second vocabulary associated with the second modality using the multimodal vocabulary mapping.

Abstract: An intelligent imaging system, includes an image generator that projects multiple angle views of a user, a plurality of cameras for capturing a plurality of images of the user, an image processing unit, a style advisor, and a control mechanism.

Abstract: The present invention is directed to a method and apparatus for assisting in rating and filtering multimedia content, such as images, videos and sound recordings. One embodiment comprises a computer implemented method for rating the objectionability of specified digital content that comprises one or more discrete content items, wherein the method includes the step of moving the specified content to one or more filtering stages in a succession of filtering stages. After the specified content is moved to a given one of the filtering stages, a rating procedure is carried out to determine whether a rating can be applied to one or more of the content items, and if so, a selected rating is applied to each of the one or more content items. The method further comprises moving content items of the specified content to the next stage in the succession after the given stage, when at least one content item of the specified content remains without rating, after the rating procedure at the given stage.

Abstract: A computer implemented method, apparatus, and computer program product code for temporal, event-based video fingerprinting. In one embodiment, events in video content are detected. The video content comprises a plurality of video frames. An event represents discrete points of interest in the video content. A set of temporal, event-based segments are generated using the events. Each temporal, event-based segment is a segment of the video content covering a set of events. A time series signal is derived from each temporal, event-based segment using temporal tracking of content-based features of a set of frames associated with the each temporal, event-based segment. A temporal segment based fingerprint is extracted based on the time series signal for the each temporal, event-based segment to form a set of temporal segment based fingerprints associated with the video content.

Abstract: A method and apparatus is disclosed that receives an input narrative in a particular media format, and produces therefrom a multimedia enriched version of the input narrative. In one embodiment, a method is provided for enriching an input narrative with multimedia content, wherein the method includes the step of selectively segmenting the input narrative to form queries pertaining to different portions or elements of the input narrative. The method further comprises retrieving multimedia artifacts from one or more multimedia repositories, wherein each retrieved multimedia artifact is associated with one of the queries, and selecting a subset of the retrieved multimedia artifacts, wherein the subset includes artifacts comprising different types of multimedia content. The input narrative is then combined with respective multimedia artifacts of the subset, in order to provide an enriched and enhanced narrative.

Abstract: A computer implemented method, apparatus, and computer usable program code for multimodal cross-vocabulary mapping. A corpus of multimodal content is annotated simultaneously using annotations from a plurality of vocabularies to form a set of common annotations. Relationships between a first vocabulary associated with a first modality and a second vocabulary associated with a second modality are identified using the set of common annotations to form a multimodal vocabulary mapping. Items in the first vocabulary associated with the first modality are mapped to items in the second vocabulary associated with the second modality using the multimodal vocabulary mapping.

Abstract: An intelligent imaging system, includes an image generator that projects multiple angle views of a user, a plurality of cameras for capturing a plurality of images of the user, an image processing unit, a style advisor, and a control mechanism.

Abstract: A technique for executing a statement in a computer to manipulate data in a data store connected to the computer. Initially, a statement having an aggregate predicate is received. Then, when evaluating the statement, the aggregate predicate is processed to rank data. Furthermore, evaluating includes performing an extended sort operator with early termination.

Abstract: A visual query system, and associated method and computer program product enhance and accelerate content-based querying, and present a new image similarity measure using known or available software applications and hardware components of video compression systems. The present system encodes images as consecutive frames in a video sequence and uses the ratio between the file length of the compressed sequence and the original file length as a distance measure. The system considers the compression ratio to be an estimate of the entropy of the combined images, which can be used to estimate the amount of new information introduced from one image to the other.

Abstract: Methods of imaging objects based on wavelet retrieval of scenes utilize wavelet transformation of plural defined regions of a query image. By increasing the granularity of the query image to greater than one region, accurate feature vectors are obtained that allow for robust extraction of corresponding regions from a database of target images. The methods further include the use of sliding windows to decompose the query and target images into regions, and the clustering of the regions utilizing a novel similarity metric that ensures robust image matching in low response times.

Abstract: A visual query system, and associated method and computer program product enhance and accelerate content-based querying, and present a new image similarity measure using known or available software applications and hardware components of video compression systems. The present system encodes images as consecutive frames in a video sequence and uses the ratio between the file length of the compressed sequence and the original file length as a distance measure. The system considers the compression ratio to be an estimate of the entropy of the combined images, which can be used to estimate the amount of new information introduced from one image to the other.

Abstract: Disclosed is a method and apparatus for querying stored multimedia data in a computer system, comprising receiving into an intermediate level a high-level concept from a user describing data to be retrieved, translating, in said intermediate level, said high-level concept into low-level queries by using system pre-defined high-level concepts, transferring said low-level queries to a low level comprising one or more search engines; said one or more search engines performing a query of the stored multimedia information using said low-level queries. A matching algorithm for use with the invention is also disclosed.