---
title: DA3Mono-LARGE-MLX
canonical_url: "https://www.modelscope.cn/models/mlx-community/DA3Mono-LARGE-MLX"
md_url: "https://www.modelscope.cn/models/mlx-community/DA3Mono-LARGE-MLX.md"
repository: mlx-community/DA3Mono-LARGE-MLX
last_updated: 2026-08-31
license: cc-by-nc-4.0
base_model:
  - depth-anything/DA3MONO-LARGE
base_model_relation: finetune
parameters: 334.1M
tensor_type:
  - F32
library_name:
  - mlx
  - safetensors
  - pytorch
frameworks:
  - pytorch
downloads: 4
stars: 0
tags:
  - mlx
  - depth-estimation
  - depth-anything
  - monocular-depth
---

# DA3Mono-LARGE-MLX

> DA3Mono-LARGE-MLX - mlx-community 在 ModelScope 开源的模型。depth-anything/DA3MONO-LARGE (Depth Anything 3 Mono, ByteDance) converted to MLX for Apple silicon.

mlx-community/DA3Mono-LARGE-MLX 是 ModelScope 魔搭社区上的 334.1M 参数机器学习模型，采用 cc-by-nc-4.0 许可，基于 depth-anything/DA3MONO-LARGE 构建。

- **Repository**: mlx-community/DA3Mono-LARGE-MLX
- **License**: cc-by-nc-4.0
- **Parameters**: 334.1M
- **Base model**: depth-anything/DA3MONO-LARGE
- **Tags**: mlx, depth-estimation, depth-anything, monocular-depth
- **Downloads**: 4
- **Stars**: 0
- **Last updated**: 2026-08-31

Source: https://www.modelscope.cn/models/mlx-community/DA3Mono-LARGE-MLX

---

# DA3Mono-LARGE-MLX

[depth-anything/DA3MONO-LARGE](https://huggingface.co/depth-anything/DA3MONO-LARGE) (Depth Anything 3 Mono, ByteDance) converted to MLX for Apple silicon.

Single-image monocular depth: DinoV2 ViT-L/14 backbone + DPT head with an auxiliary sky-logit branch. Positional embedding is pre-baked (bicubic) for a fixed 504×504 input.

- **Input**: `(1, 3, 504, 504)` float32, ImageNet-normalized (`(x/255 - mean) / std`)
- **Outputs**: `depth (504, 504)` = `exp(logits)`, `sky (504, 504)` = `relu(logits)`
- **Precision**: fp32 weights (see [DA3Mono-LARGE-MLX-fp16](https://huggingface.co/Jup33QE/DA3Mono-LARGE-MLX-fp16) for the fp16 variant)

## Use with MLX

```python
import mlx.core as mx
import numpy as np
from PIL import Image
from da3mono_mlx import DA3MonoMLX   # shipped in this repo

model = DA3MonoMLX().load_weights("model.safetensors")

img = Image.open("photo.jpg").convert("RGB").resize((504, 504), Image.BICUBIC)
a = np.asarray(img).astype(np.float32) / 255.0
x = (a - [0.485, 0.456, 0.406]) / [0.229, 0.224, 0.225]
x = mx.array(x.transpose(2, 0, 1)[None])          # NCHW

depth, sky = model(x)
mx.eval(depth, sky)
```

## Accuracy vs. PyTorch reference

Validated against the official PyTorch forward on both random tensors and real images:

| metric | value |
| --- | --- |
| depth Pearson r | 0.9999998 |
| depth median relative error | 0.07 % |
| depth max abs error | 0.02 |

## License & citation

Weights are a conversion of `depth-anything/DA3MONO-LARGE`; the original
model is released under CC-BY-NC-4.0 (non-commercial). Please cite:

```bibtex
@article{depthanything3,
  title={Depth Anything 3: Recovering the Visual Space from Any Views},
  author={Depth Anything Team},
  journal={arXiv preprint arXiv:2511.10647},
  year={2025}
}
```