Object modeling based on frames of a video

Is it possible to do object-based modeling of frames in a video? In the video appear objects (for example a cube, cone ....) and as the camera films, the video gives different perspectives of the objects, what I want to know is if it is possible to project this object directly in a 3d modeler (blender ...) or better if it is possible to load all frames of the video so that you can create a 3d scene based on the objects of the video