The vvas_xmetaaffixer plug-in is designed to adjust the metadata information for different resolutions. When performing machine learning on a frame with a different resolution or color format from the original, the metadata can still be associated with the original, full-resolution frame. The vvas_metaaffixer includes two types of pads: master pads and slave pads. The number of input pads created depends on the number of input sources and can be requested as needed. The plug-in requires at least one mandatory master input pad (sink pad) to receive the original/reference metadata, while the remaining input pads are referred to as slave pads. The metadata received on the master sink pad is adjusted based on the resolution of each slave sink pad. The adjusted metadata is then added to the output buffer coming out of the source slave pads. Up to 16 slave pads can be created as required.
For insight into the implementation details, refer to the vvas_xmetaaffixer source code located at <vvas>/vvas-gst-plugins/gst/metaaffixer/.