yzcjtr / GeoNet

Code for GeoNet: Unsupervised Learning of Dense Depth, Optical Flow and Camera Pose (CVPR 2018)
MIT License
726 stars 181 forks source link

How are target and source frames selected? #71

Closed qixxyy closed 3 years ago

qixxyy commented 3 years ago

Does the target frame refer to the first image in the sequence or the middle image? If the three pictures in the sequence are divided into T-1, T, and T +1, then forward refers to the flow between T-1 and T, and backward refers to the flow between T and T +1, right?

qixxyy commented 3 years ago

yes