Motion compensation using temporal picture interpolation

An apparatus for generating enhanced reference pictures in a video encoding or decoding process, comprising:

• at least one processor;
• at least one memory including computer program code which, when executed by the at least one processor, causes the apparatus to:
• receive one or more known reference elements of video data from a reference picture buffer, wherein the one or more known references elements include part of a hierarchical group of pictures (GOP);
• generate, using a neural network, one or more additional reference elements of video data from the one or more known reference elements of video data, wherein the neural network has been trained on known sets of reference pictures in hierarchical GOPs to recreate pictures in higher hierarchical layers of the hierarchical GOPs from pictures in lower layers of the hierarchical GOPs;
• and output, from the neural network, the one or more additional reference elements of video data;
• wherein generating the one or more additional reference elements of video data from the one or more known reference elements of video data comprises use of temporal interpolation.

A computer readable medium having computer readable code stored thereon for generating enhanced reference pictures in a video encoding or decoding process, the computer readable code, when executed by at least one processor, cause the at least one processor to:

• receive one or more known reference elements of video data from a reference picture buffer;
• generate, using one or more hierarchical algorithms, one or more additional reference elements of video data from the one or more known reference elements of video data, wherein the one or more hierarchical algorithms are trained on known sets of reference pictures in hierarchical Groups of Pictures (GOPs) or previously decoded pictures;
• and
• output the one or more additional reference elements of video data;
• wherein generating the one or more additional reference elements of video data from the one or more known reference elements of video data comprises use of temporal interpolation.