mlx-community/rfdetr-seg-xlarge-fp32

This model was converted to MLX format from RF-DETR (ICLR 2026) using mlx-vlm version 0.4.3.

Use with mlx

pip install -U mlx-vlm

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

result = predictor.predict(Image.open("image.jpg"))

CLI

python -m mlx_vlm.models.rfdetr.generate --task segment --image photo.jpg --model mlx-community/rfdetr-seg-xlarge-fp32
python -m mlx_vlm.models.rfdetr.generate --task track --video input.mp4 --model mlx-community/rfdetr-seg-xlarge-fp32
python -m mlx_vlm.models.rfdetr.generate --task realtime --model mlx-community/rfdetr-seg-xlarge-fp32

Model Details

Architecture	DINOv2-small + C2f + 6-layer decoder + 6-block seg head
Task	Object detection + instance segmentation (COCO 80 classes)
Parameters	~38M
Input resolution	624x624
Mask resolution	156x156
Dtype	float32

All RF-DETR MLX Models

Model	Resolution	Masks	Use case
rfdetr-base-fp32	560px	No	Fast detection
rfdetr-seg-small-fp32	384px	96x96	Realtime segmentation
rfdetr-seg-large-fp32	504px	126x126	Better masks
rfdetr-seg-xlarge-fp32	624px	156x156	High quality
rfdetr-seg-2xlarge-fp32	768px	192x192	Best quality

Reference

• RF-DETR (ICLR 2026)
• Roboflow RF-DETR