Contextual in-game element recognition, annotation and interaction based on remote user input

Systems, methods, and apparatuses are provided for annotating a video frame generated by a video game. A video game model that associates element tags with elements of the video game may be generated. The video game model may be applied by a video game overlay executing concurrently with the video game. The video game overlay may receive a remote user input from one or more remote devices over a network. The remote user input may be multiplexed and/or normalized, and subsequently parsed by applying the video game model to extract an element tag corresponding to the video game. By applying the video game model, an in-game element of the video game corresponding to the element tag may be identified in the video frame. Based on the identified element in the video frame of the video game, the video frame may be annotated and presented to the video game user.