---
title: rfdetr-base-fp32
canonical_url: "https://www.modelscope.cn/models/mlx-community/rfdetr-base-fp32"
md_url: "https://www.modelscope.cn/models/mlx-community/rfdetr-base-fp32.md"
repository: mlx-community/rfdetr-base-fp32
last_updated: 2026-03-31
model_type:
  - rf-detr
base_model:
  - roboflow/rf-detr
base_model_relation: finetune
parameters: 32.2M
tensor_type:
  - F32
library_name:
  - mlx
  - safetensors
  - pytorch
frameworks:
  - pytorch
language:
  - en
downloads: 7
stars: 0
tags:
  - mlx
  - rf-detr
  - object-detection
  - apple-silicon
---

# rfdetr-base-fp32

> rfdetr-base-fp32 - mlx-community 在 ModelScope 开源的模型。mlx-community/rfdetr-base-fp32

mlx-community/rfdetr-base-fp32 是 ModelScope 魔搭社区上的 32.2M 参数机器学习模型，基于 roboflow/rf-detr 构建。

- **Repository**: mlx-community/rfdetr-base-fp32
- **Parameters**: 32.2M
- **Base model**: roboflow/rf-detr
- **Tags**: mlx, rf-detr, object-detection, apple-silicon
- **Downloads**: 7
- **Stars**: 0
- **Last updated**: 2026-03-31

Source: https://www.modelscope.cn/models/mlx-community/rfdetr-base-fp32

---

# mlx-community/rfdetr-base-fp32

This model was converted to MLX format from [RF-DETR](https://github.com/roboflow/rf-detr) (ICLR 2026) using [mlx-vlm](https://github.com/Blaizzy/mlx-vlm) version **0.4.3**.

RF-DETR is a real-time detection transformer achieving state-of-the-art performance on COCO.

## Use with mlx

```bash
pip install -U mlx-vlm
```

```python
from pathlib import Path
from PIL import Image
from mlx_vlm.utils import load_model
from mlx_vlm.models.rfdetr.processing_rfdetr import RFDETRProcessor
from mlx_vlm.models.rfdetr.generate import RFDETRPredictor

model = load_model(Path("mlx-community/rfdetr-base-fp32"))
processor = RFDETRProcessor.from_pretrained("mlx-community/rfdetr-base-fp32")
predictor = RFDETRPredictor(model, processor, score_threshold=0.3, nms_threshold=0.5)

result = predictor.predict(Image.open("image.jpg"))
for name, score, box in zip(result.class_names, result.scores, result.boxes):
    print(f"{name}: {score:.2f} [{box[0]:.0f}, {box[1]:.0f}, {box[2]:.0f}, {box[3]:.0f}]")
```

### CLI

```bash
# Image
python -m mlx_vlm.models.rfdetr.generate --image photo.jpg --model mlx-community/rfdetr-base-fp32

# Video
python -m mlx_vlm.models.rfdetr.generate --video input.mp4 --model mlx-community/rfdetr-base-fp32

# Realtime camera
python -m mlx_vlm.models.rfdetr.generate --task realtime --model mlx-community/rfdetr-base-fp32
```

## Model Details

| | |
|---|---|
| Architecture | DINOv2-small backbone + C2f projector + Deformable DETR decoder |
| Task | Object detection (COCO 80 classes) |
| Parameters | ~32M |
| Input resolution | 560x560 |
| Dtype | float32 |
| Inference (M4 Max) | ~32ms per image (~31 FPS) |

## Reference

- [RF-DETR: Real-Time Detection Transformer](https://arxiv.org/abs/2511.09554) (ICLR 2026)
- [Roboflow RF-DETR](https://github.com/roboflow/rf-detr)