Depth Any Canopy Small

This is the small version of Depth Any Canopy presented in Depth Any Canopy Paper. A Base version is also available. a

Model Details

The model is Depth-Anything-Small finetuned for canopy height estimation on a filtered set of EarthView.

• License: Apache 2.0
• Finetuned from model: Depth-Anything-Small

Uses and Limitations

The model is capable of working with aerial imagery of NEON. The coverage is limited to the US. We cannot guarantee its generalizability over other areas of the globe. The images cover only RGB channels; no study of hyperspectral imagery was done.

How to Get Started with the Model

Use the code below to get started with the model.

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("depth-estimation", model="DarthReca/depth-any-canopy-base")

# Load model directly
from transformers import AutoModelForDepthEstimation

model = AutoModelForDepthEstimation.from_pretrained("DarthReca/depth-any-canopy-base")

Environmental Impact

• Carbon Emitted: 0.14 kgCO2

Carbon emissions are estimated using the Machine Learning Impact calculator.

Citation

BibTeX:

@inbook{RegeCambrin2025,
  title = {Depth Any Canopy: Leveraging Depth Foundation Models for Canopy Height Estimation},
  ISBN = {9783031923876},
  ISSN = {1611-3349},
  url = {http://dx.doi.org/10.1007/978-3-031-92387-6_5},
  DOI = {10.1007/978-3-031-92387-6_5},
  booktitle = {Computer Vision – ECCV 2024 Workshops},
  publisher = {Springer Nature Switzerland},
  author = {Rege Cambrin,  Daniele and Corley,  Isaac and Garza,  Paolo},
  year = {2025},
  pages = {71–86}
}