# D-FINE-X

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

- Family: D-FINE (USTC)
- Parameters: 62.0M
- GFLOPs: 202.0
- Default input size: 640px
- Detection: detr · nms-free
- Architecture: transformer · HGNetv2 backbone
- Weight license: Apache-2.0
- Paper-reported mAP@50-95: 55.8%
- Paper: https://arxiv.org/abs/2410.13842
- Original code: https://github.com/Peterande/D-FINE

## Benchmarks (COCO val2017)

| Hardware | Runtime | mAP@50-95 | FPS | Latency (ms) | VRAM (MB) |
|---|---|---|---|---|---|
| NVIDIA A100 | PyTorch FP32 | 59.3 | 15.4 | 64.7 | 401 |
| NVIDIA Jetson Orin Nano Super 8GB | ONNX Runtime FP32 | 61.4 | 3.0 | 328.9 | - |
| NVIDIA Jetson Orin Nano Super 8GB | PyTorch FP32 | 61.4 | 2.9 | 350.0 | 399 |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP16 | 61.4 | 14.2 | 70.5 | - |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP32 | 61.5 | 8.0 | 124.8 | - |
| NVIDIA RTX 5070 Ti | ONNX Runtime FP32 | 61.4 | 41.5 | 24.1 | - |
| NVIDIA RTX 5070 Ti | PyTorch FP32 | 61.4 | 18.1 | 55.1 | 404 |
| NVIDIA RTX 5070 Ti | TensorRT FP16 | 61.5 | 73.0 | 13.7 | - |
| NVIDIA RTX 5070 Ti | TensorRT FP32 | 61.4 | 57.9 | 17.3 | - |

## Usage with LibreYOLO

```python
from libreyolo import LibreYOLO

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

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