Video interpretation apparatus and method

A video interpretation method performed by a video interpretation apparatus, comprising:

• generating object information based on objects in an input video;
• generating a dynamic spatial relation between objects based on the object information;
• generating general event information based on the dynamic spatial relation;
• and
• generating video information including any one of a sentence and an event description based on the object information and the general event information,
• wherein the object information includes an ID and a spatial region of each of the objects for a specific frame in a video,
• wherein the spatial region is represented by a Minimum Bounding Polygon (MBP) including each of the objects, for the specific frame, and
• wherein the dynamic spatial relation includes a relation type which is based on variation in a spatial relation between the objects, time interval information, and a spatial region of a relation.

A video interpretation method performed by a video interpretation apparatus, comprising:

• generating object information based on objects in an input video;
• generating a dynamic spatial relation between objects based on the object information;
• generating general event information based on the dynamic spatial relation;
• and
• generating video information including any one of a sentence and an event description based on the object information and the general event information,
• wherein generating the general event information comprises:
• selecting any one from among prepositions and adverbs corresponding to the dynamic spatial relation;
• selecting only visually representable specific meanings, from among specific meanings corresponding to the prepositions and the adverbs;
• generating relation IDs respectively corresponding to the visually representable specific meanings;
• and
• generating the general event information based on the object information corresponding to the relation IDs.