# RF-DETR-S

> RF-DETR-S: transformer object detector from the RF-DETR family. Available in LibreYOLO, the MIT-licensed open-source library (free for commercial use).

- Family: RF-DETR (Roboflow)
- Parameters: 32.1M
- GFLOPs: 0.0
- Default input size: 512px
- Detection: detr · nms-free
- Architecture: transformer · DINOv2 backbone
- Weight license: Apache-2.0
- Paper-reported mAP@50-95: 53%
- Paper: https://blog.roboflow.com/rf-detr
- Original code: https://github.com/roboflow/rf-detr

## Benchmarks (COCO val2017)

| Hardware | Runtime | mAP@50-95 | FPS | Latency (ms) | VRAM (MB) |
|---|---|---|---|---|---|
| NVIDIA Jetson Orin Nano Super 8GB | ONNX Runtime FP32 | 55.1 | 6.2 | 161.5 | - |
| NVIDIA Jetson Orin Nano Super 8GB | PyTorch FP32 | 55.1 | 6.9 | 144.6 | 175 |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP16 | 55.2 | 15.5 | 64.4 | - |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP32 | 55.1 | 14.5 | 69.1 | - |
| NVIDIA RTX 5070 Ti | ONNX Runtime FP32 | 55.1 | 68.5 | 14.6 | - |
| NVIDIA RTX 5070 Ti | PyTorch FP32 | 55.1 | 35.8 | 27.9 | 176 |
| NVIDIA RTX 5070 Ti | TensorRT FP16 | 55.3 | 71.6 | 14.0 | - |
| NVIDIA RTX 5070 Ti | TensorRT FP32 | 55.1 | 77.5 | 12.9 | - |
| Raspberry Pi 5 | ONNX Runtime FP32 | 55.1 | 1.2 | 850.1 | - |
| Raspberry Pi 5 | PyTorch FP32 | 55.1 | 1.0 | 959.4 | - |

## Usage with LibreYOLO

```python
from libreyolo import LibreYOLO

model = LibreYOLO("LibreRFDETRs.pt")
result = model("image.jpg", conf=0.25, iou=0.45)
```

Source: https://www.visionanalysis.org/model/rfdetr-s. Benchmarks produced with LibreYOLO (https://github.com/Libre-YOLO/libreyolo).