monodepth2

Monodepth2 is a PyTorch implementation designed for self-supervised monocular depth estimation, allowing users to train and test models based on recent research in depth prediction. Its primary use case includes generating depth maps from single images or stereo images, with multiple pretrained model configurations available for different resolution and training modalities. Notable features include support for varying input types, easy setup instructions, and functionality for predicting both scaled disparity and metric depth.