---
title: dinov3-vitl16-chmv2-dpt-head
canonical_url: "https://www.modelscope.cn/models/facebook/dinov3-vitl16-chmv2-dpt-head"
md_url: "https://www.modelscope.cn/models/facebook/dinov3-vitl16-chmv2-dpt-head.md"
repository: facebook/dinov3-vitl16-chmv2-dpt-head
last_updated: 2026-07-18
model_type:
  - chmv2
architectures:
  - CHMv2ForDepthEstimation
base_model:
  - facebook/dinov3-vitl16-chmv2-dpt-head
base_model_relation: finetune
parameters: 336.9M
tensor_type:
  - F32
library_name:
  - pytorch
  - transformer
  - safetensors
frameworks:
  - pytorch
language:
  - en
downloads: 286
stars: 0
tags:
  - chmv2
  - dinov3
---

# dinov3-vitl16-chmv2-dpt-head

> dinov3-vitl16-chmv2-dpt-head - facebook 在 ModelScope 开源的模型。Model Card for CHMv2

facebook/dinov3-vitl16-chmv2-dpt-head 是 ModelScope 魔搭社区上的 336.9M 参数机器学习模型，基于 facebook/dinov3-vitl16-chmv2-dpt-head 构建。

- **Repository**: facebook/dinov3-vitl16-chmv2-dpt-head
- **Parameters**: 336.9M
- **Base model**: facebook/dinov3-vitl16-chmv2-dpt-head
- **Tags**: chmv2, dinov3
- **Downloads**: 286
- **Stars**: 0
- **Last updated**: 2026-07-18

Source: https://www.modelscope.cn/models/facebook/dinov3-vitl16-chmv2-dpt-head

---

# Model Card for CHMv2

The Canopy Height Maps v2 (CHMv2) model is a DPT-based decoder estimating canopy height given satellite imagery, leveraging DINOv3 as the backbone. Building on our original high-resolution canopy height maps released in 2024, CHMv2 delivers substantial improvements in accuracy, detail, and global consistency.

## Model Details

CHMv2 model was developed using the satellite DINOv3 ViT-L as the frozen backbone. Released with world-scale maps generated with it, they will help researchers and governments measure and understand every tree, gap, and canopy edge — enabling smarter biodiversity support and land-management decisions.

## Usage With Transformers

Run inference on an image with the following code:

```python
from PIL import Image
import torch

from transformers import CHMv2ForDepthEstimation, CHMv2ImageProcessorFast

processor = CHMv2ImageProcessorFast.from_pretrained("facebook/dinov3-vitl16-chmv2-dpt-head")
model = CHMv2ForDepthEstimation.from_pretrained("facebook/dinov3-vitl16-chmv2-dpt-head")

image = Image.open("image.tif")
inputs = processor(images=image, return_tensors="pt")

with torch.no_grad():
    outputs = model(**inputs)

depth = processor.post_process_depth_estimation(
    outputs, target_sizes=[(image.height, image.width)]
)[0]["predicted_depth"]
```

### Model Description

- **Developed by:** Meta AI
- **Model type:** DPT head
- **License:** [DINOv3 License](https://ai.meta.com/resources/models-and-libraries/dinov3-license/)

### Model Sources

- **Repository:** [https://github.com/facebookresearch/dinov3](https://github.com/facebookresearch/dinov3)
- **Paper:** [https://arxiv.org/abs/2603.06382](https://arxiv.org/abs/2603.06382)


### Direct Use

The model can be used without fine-tuning to obtain competitive results on various satellite datasets ([paper link](https://arxiv.org/abs/2603.06382)).