lppllppl920/EndoscopyDepthEstimation-Pytorch

In terms of the format, please refer to one training data example in this repository. We use SfM to generate training data in this work. Color images with the format of "{:08d}.jpg" are extracted from the video sequence where SfM is applied. In this example, since all images are from the same video sequence, we assume the intrinsic matrices are the same for all images. The first three rows in this file are focal length, x and y of the principal point of the camera of the first image.