Method and system for using semantic-segmentation for automatically generating effects and transitions in video productions

A method of automatic video production comprising:

• obtaining a set of media entities comprising pixels, wherein at least one of the media entities comprises a background and at least one foreground object;
• automatically analyzing the media entities using visual analysis;
• computing, for at least one of said media entities, semantic segmentation indicative of a support of the at least one foreground object, based on said visual analysis, wherein said computing assigns each of said pixels with respective pixel values based on the sematic segmentation;
• generating at least one transition being temporally ordered between at least two of the media entities, wherein a transition between a first media entity to a second media entity exhibits a combination of the first and the second media entities, and in which, for at least one of the media entities, the pixels belonging to a foreground object and the pixels belonging to the background undergo two different visual operations, wherein in a time prior to the transition, only said first visual portion is visible and wherein in a time post to said transition only said second visual portion is visible, wherein the transition is generated based on said respective pixel values;
• and
• generating a video production by combining a plurality of the media entities into one video production, while including the said at least one transition in the video production.

A method of automatic video production comprising:

• obtaining a set of media entities comprising pixels, wherein at least one of the media entities comprises a background and at least one foreground object;
• automatically analyzing the media entities using visual analysis;
• automatically selecting at least two visual portions, based on said visual analysis;
• computing, for at least one of said visual portions, semantic segmentation indicative of a support of the at least one foreground object, based on said visual analysis, wherein said computing assigns said pixels with respective pixel values based on the sematic segmentation;
• generating at least one visual effect in which the foreground object and the background undergo two different visual operations, wherein in a time prior to applying the visual effect, only one of said visual portions is visible and wherein in a time post to said applying of the visual effect only another one of said visual portions is visible, wherein the visual effect is generated based on said respective pixel values;
• and
• generating a video production by combining a plurality of the visual portions into one video production, while including said at least one visual effect in the video production.

A non-transitory computer readable medium comprising a set of instructions that, when executed, cause at least one computer processor to:

• obtain a set of media entities comprising pixels, wherein at least one of the media entities comprises a background and at least one foreground object;
• automatically analyze the media entities using visual analysis;
• automatically select at least two visual portions, based on said visual analysis;
• compute, for at least one of said visual portions, semantic segmentation indicative of a support of the at least one foreground object, based on said visual analysis;
• assign each of said pixels with respective pixel values based on the sematic segmentation;
• generate at least one masking transition being temporally ordered between at least two of the visual portions, wherein a masking transition between a first visual portion to a second visual portion exhibits a combination of the first and the second visual portions, in which the foreground object of the first visual portion appears over the second visual portion or the foreground object of the second visual portion appears over the first visual portion, wherein in a time prior to the transition, only said first visual portion is visible and wherein in a time post to said transition only said second visual portion is visible, wherein the masking transition is generated based on said respective pixel values;
• and
• generate a video production by combining a plurality of the visual portions into one video production, while including the at least one masking transition in the video production in said temporal order.