Greater prediction accuracy during inference

[jmayank23]

Thank you so much for this fantastic repo! If I were to calculate the camera intrinsics for the camera I use for my input video, can I expect higher accuracy predictions, i.e., 3D reconstruction? If so, where can I update the camera intrinsics?