Daniele Rege Cambrin1 · Isaac Corley2 · Paolo Garza1
1Politecnico di Torino, Italy 2University of Texas at San Antonio, USA
In this paper, we propose transferring the representations learned by recent depth estimation foundation models to the remote sensing domain for measuring canopy height. Our findings suggest that our proposed Depth Any Canopy, the result of fine-tuning the Depth Anything v2 model for canopy height estimation, provides a performant and efficient solution, surpassing the current state-of-the-art with superior or comparable performance using only a fraction of the computational resources and parameters. Furthermore, our approach requires less than $1.30 in compute and results in an estimated carbon footprint of 0.14 kgCO2.
REPOSITORY IN CONSTRUCTION SOME FILES COULD BE MISSING
Install the dependencies of the requirements.txt file. Make sure to edit the config files in the configs/
folder. Then simply run main.py
Pre-trained checkpoints are available on HuggingFace.
Model | Parameters | Checkpoint |
---|---|---|
Depth-Any-Canopy-Small | 24.8M | Download |
Depth-Any-Canopy-Base | 97.5M | Download |
You can easily load them with pipelines or AutoModel:
# 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")
This project is licensed under the Apache 2.0 license. See LICENSE for more information.
If you find this project useful, please consider citing:
@misc{cambrin2024depthcanopyleveragingdepth,
title={Depth Any Canopy: Leveraging Depth Foundation Models for Canopy Height Estimation},
author={Daniele Rege Cambrin and Isaac Corley and Paolo Garza},
year={2024},
eprint={2408.04523},
archivePrefix={arXiv},
primaryClass={cs.CV},
url={https://arxiv.org/abs/2408.04523},
}