Context--
CategoryVision Model
About RF-DETR
RF-DETR is Roboflow’s real-time transformer model family for object detection, instance segmentation, and preview keypoint detection. The source README reports Microsoft COCO and RF100-VL benchmarks and public Apache-licensed model variants.
Capabilities
object detectioninstance segmentationkeypoint detectionvisionfine tuning
Input Modalities
image
Output Modalities
bounding-boxesmaskskeypoints
Technical Details
- API Identifier
- https://github.com/roboflow/rf-detr
- Category
- Vision Model
Tags
computer-visionobject-detectionsegmentationopen-sourceroboflowgithubfine-tuning
Benchmarks
Performance scores for RF-DETR across standard benchmarks.
COCO AP50:95 Object DetectionRoboflow RF-DETR README benchmark table · Jun 2026
60.1
RF100-VL AP50:95 Object DetectionRoboflow RF-DETR README benchmark table · Jun 2026
63.2
COCO Keypoint AP50:95Roboflow RF-DETR README benchmark table · Jun 2026
71.8
Pricing and access
No verified hosted per-token price is available for this repository listing. Self-hosted inference has hardware and infrastructure costs; deployment and licensing vary by checkpoint.
Roboflowsaas
Computer vision tools for developers and enterprises
Founded 2019United States
View full profile