System and method for video selection and labelling

A system for video selection and labelling comprising:

• a computer memory arranged to receive a first video assembly comprising a sequence of video frames and a plurality of data items;
• an electronic display configured to (a) present to a human operator, the first video assembly;
• and (b) present to the human operator the one or more candidate video assemblies;
• a user interface, associated with the electronic display, arranged to (a) receive from the human operator a selection of one or more first data items from the plurality of data items within the first video assembly;
• and (b) receive feedback from the human operator, over the user interface, whether the label generated for the one or more candidate video assemblies is acceptable;
• and
• a computer processor arranged to identify one or more second video assemblies comprising a sequence of frames and one or more data items, wherein the one or more second data items are similar to the selection of the one or more first data items, wherein the computer processor is arranged to generate one or more candidate video assemblies each comprising:
• a sequence of frames, wherein the sequence of frames comprises a subset of the frames and data items from the one or more second video assemblies which have been identified to be similar to the one or more first data items, and a natural language label describing the sequence of frames.

A non-transitory computer readable medium for video selection and labelling comprising a

• set of instructions that, when executed, cause at least one computer processor to:
• receive a first video assembly comprising a sequence of frames and a plurality of data items;
• present a human operator, over an electronic display, the first video assembly;
• receive from the human operator, over a user interface associated with the electronic display, a selection of one or more first data items from the plurality of data items within the first video assembly;
• identify one or more second video assemblies comprising a sequence of frames and one or more second data items, wherein the one or more second data items are similar to the selection of the one or more first data items;
• generate one or more candidate video assemblies each comprising:
• a sequence of frames, wherein the sequence of frames comprises a subset of the frames and data items from the one or more second video assemblies which have been identified to be similar to the one or more first data items, and a natural language label describing the sequence of frames;
• present the human operator, over the electronic display, the one or more candidate video assemblies;
• and
• receive feedback from the human operator, whether the label generated for the one or more candidate video assemblies is acceptable.