# RT-DETR-R18

> RT-DETR-R18: transformer object detector from the RT-DETR family. Available in LibreYOLO, the MIT-licensed open-source library (free for commercial use).

- Family: RT-DETR (Baidu)
- Parameters: 20.0M
- GFLOPs: 60.0
- Default input size: 640px
- Detection: detr · nms-free
- Architecture: transformer · ResNet-18 backbone
- Weight license: Apache-2.0
- Paper-reported mAP@50-95: 46.5%
- Paper: https://arxiv.org/abs/2304.08069
- Original code: https://github.com/lyuwenyu/RT-DETR

## Benchmarks (COCO val2017)

| Hardware | Runtime | mAP@50-95 | FPS | Latency (ms) | VRAM (MB) |
|---|---|---|---|---|---|
| NVIDIA A100 | PyTorch FP32 | 45.6 | 27.4 | 36.5 | 155 |
| NVIDIA Jetson Orin Nano Super 8GB | ONNX Runtime FP32 | 49.8 | 9.6 | 103.7 | - |
| NVIDIA Jetson Orin Nano Super 8GB | PyTorch FP32 | 49.8 | 10.2 | 98.3 | 155 |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP16 | 49.7 | 38.8 | 25.7 | - |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP32 | 49.8 | 22.1 | 45.3 | - |
| NVIDIA RTX 5070 Ti | ONNX Runtime FP32 | 49.8 | 92.5 | 10.8 | - |
| NVIDIA RTX 5070 Ti | PyTorch FP32 | 49.8 | 47.2 | 21.2 | 169 |
| NVIDIA RTX 5070 Ti | TensorRT FP16 | 49.8 | 113.4 | 8.8 | - |
| NVIDIA RTX 5070 Ti | TensorRT FP32 | 49.8 | 120.6 | 8.3 | - |

## Usage with LibreYOLO

```python
from libreyolo import LibreYOLO

model = LibreYOLO("LibreRTDETRr18.pt")
result = model("image.jpg", conf=0.25, iou=0.45)
```

Source: https://www.visionanalysis.org/model/rtdetr-r18. Benchmarks produced with LibreYOLO (https://github.com/Libre-YOLO/libreyolo).