EPFL-VILAB / omnidata

A Scalable Pipeline for Making Steerable Multi-Task Mid-Level Vision Datasets from 3D Scans [ICCV 2021]
Other
395 stars 49 forks source link

Checkpoint for DPT-Large for depth estimation #57

Open adondera opened 1 year ago

adondera commented 1 year ago

Hey,

The README on your torch page mentions that for depth estimation you provide two v2 models (one for the DPT-Hybrid, one for the DPT-Large architecture). However, the download script only downloads the DPT-Hybrid model for versions 1 and 2.

Would it also be possible to also make the DPT-Large model available?

Thanks!