search

DepthAnything / Depth-Anything-V2

[NeurIPS 2024] Depth Anything V2. A More Capable Foundation Model for Monocular Depth Estimation
https://depth-anything-v2.github.io
Apache License 2.0
3.99k stars 345 forks source link

Question about Metric Depth Training #167

Open shilpaullas97 opened 2 months ago

shilpaullas97 commented 2 months ago

Hi @LiheYoung ,

Thank you so much for sharing this work. I'm new to this monocular depth estimation tasks. I would like to clarify few doubts.

  1. Is the metric depth training given in https://github.com/DepthAnything/Depth-Anything-V2/tree/main/metric_depth uses Teacher-Student architecture? If yes, how can we use it for our training?
  2. What are the steps to be followed while training a metric depth model for custom dataset(preferably outdoor)
  3. Are the metric depth models (example , metric_vitl - https://huggingface.co/depth-anything/Depth-Anything-V2-Metric-VKITTI-Large/resolve/main/depth_anything_v2_metric_vkitti_vitl.pth?download=true) given in the repo trained using the same script provided in metric depth folder(https://github.com/DepthAnything/Depth-Anything-V2/blob/main/metric_depth/train.py)?