Hi all, as the title said, I want to put virtual object (e.g. A cup) on the desktop in a video, the video is downloaded from the web, so I don't know the camera parameters. I tried to use mono-slam (ORBSLAM2), but without camera calibration data, I cannot render object in the video correctly, and mono-slam algorithms need initialization, while I want to render the object from start. So is there another way to achieve this ?
