Tracking of a dynamic container in a video

Tracking of a defined area over a series of video frames that compose a video is provided. Various rules regulate the identification of a reference area to be tracked and the determination of subsequent areas that correspond to the reference areas over a series of respective images included the series of video frames. Metadata indicative of placement of a subsequent area in a subsequent image can be updated and retained. A mask video that can permit avoiding or otherwise mitigating occlusion of foreground elements of the subsequent image also can be updated and retained. Such metadata and mask video can permit injecting customized content into the video for which the reference area is tracked.