Video-generating systems and methods
AssigneeShutterstock, Inc.
InventorYuriy Chemolosov
Life of this patent
Estimating the term…
Abstract
Various aspects of the subject technology relate to systems, methods, and machine-readable media for video generation. An input video can be provided to a video server that is coupled to a video database storing hundreds, thousands, millions, or billions of video clips. Each stored video clip includes content that is known to the video server. The video server identifies aspects of the content of the input video and identifies matching video clips stored in the video database, using a trained machine-learning engine. The machine-learning engine can also modify some aspects of the input video while maintaining other aspects of the input video.
Claims20
A computer-implemented method, comprising:
- receiving, from a user device, an input video;
- obtaining video from a video database using a machine-learning engine that is trained to obtain a matching video from the video database based on the input video;
- generating an output video corresponding to the input video by replacing at least a portion of the input video with the matching video;
- providing the output video to the user device;
- identifying the video with the machine-learning engine based on a filter and the input video, wherein the output video matches at least one aspect of the input video and differs from at least another aspect of the input video according to the filter;
- extracting a plurality of input clips from the input video;
- identifying, for each input clip, a corresponding video clip that is stored in the video database using the machine-learning engine;
- and
- matching the corresponding video clip in a sequence of corresponding video clips, to form a coherent scene in an output video by weighting, in the corresponding video clip, a descriptor based on a number of input clips that include the descriptor.
A system, comprising:
- a video database storing video clips;
- one or more processors;
- and
- a computer-readable storage medium coupled to the one or more processors, the computer-readable storage medium including instructions that, when executed by the one or more processors, cause the one or more processors to:
- receive, from a user device, an input video;
- obtain a video clip from the video database using a machine-learning engine that is trained to obtain a matching video from the video database based on the input video;
- generate an output video corresponding to the input video by replacing at least a portion of the input video with the video clip;
- identify, based on a filter and the input video, the output video with the machine-learning engine, wherein the output video matches at least one aspect of the input video and differs from at least another aspect of the input video according to the filter;
- extract a plurality of input clips from the input video;
- identify, for each input clip, a corresponding video clip that is stored in the video database using the machine-learning engine;
- and
- match the corresponding video clip in a sequence of corresponding video clips, to form a coherent scene in an output video by weighting, in the video clip, a descriptor based on a number of input clips that include the descriptor.
A non-transitory computer-readable medium including instructions that, when executed by a processor, cause a computer to:
- receive, from a user device, an input video;
- obtain, using a machine-learning engine and a video databases matching video from the video database that matches the input video;
- generate an output video based on a matching video;
- identify the output video based on a filter, wherein the output video matches at least one aspect of the input video and differs from a second aspect, according to the filter;
- extract a plurality of input clips;
- identify, in the video database, and for each input clip, a corresponding video clip;
- and
- match the corresponding video clip in a sequence of corresponding video clips, to form a coherent scene in an output video by weighting, in the corresponding video clip, a descriptor based on a number of input clips that include the descriptor.
Dependencies are inferred from the claim text.