The "head" refers to the component of an object detector responsible for generating predictions. It utilizes the features extracted by the "neck" (the middle processing stage) of the object detector to produce output such as bounding box coordinates and object class labels.
