cjwbw
/
depth-anything
Highly practical solution for robust monocular depth estimation by training on a combination of 1.5M labeled images and 62M+ unlabeled images
Want to make some of these yourself?
Run this model