# YOLOv9-M

> YOLOv9-M: one-stage object detector from the YOLOv9 family. Available in LibreYOLO, the MIT-licensed open-source library (free for commercial use).

- Family: YOLOv9 (Academia Sinica)
- Parameters: 20.1M
- GFLOPs: 38.7
- Default input size: 640px
- Detection: anchor-free · nms
- Architecture: one-stage · GELAN backbone
- Weight license: MIT
- Paper-reported mAP@50-95: 51.4%
- Paper: https://arxiv.org/abs/2402.13616
- Original code: https://github.com/MultimediaTechLab/YOLO

## Benchmarks (COCO val2017)

| Hardware | Runtime | mAP@50-95 | FPS | Latency (ms) | VRAM (MB) |
|---|---|---|---|---|---|
| NVIDIA Jetson Orin Nano Super 8GB | ONNX Runtime FP32 | 56.1 | 7.8 | 128.1 | - |
| NVIDIA Jetson Orin Nano Super 8GB | PyTorch FP32 | 56.1 | 8.0 | 124.8 | 183 |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP16 | 56.1 | 25.8 | 38.8 | - |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP32 | 56.1 | 14.5 | 68.9 | - |
| NVIDIA RTX 5070 Ti | ONNX Runtime FP32 | 56.1 | 64.5 | 15.5 | - |
| NVIDIA RTX 5070 Ti | PyTorch FP32 | 56.1 | 37.1 | 26.9 | 184 |
| NVIDIA RTX 5070 Ti | TensorRT FP16 | 56.1 | 72.3 | 13.8 | - |
| NVIDIA RTX 5070 Ti | TensorRT FP32 | 56.1 | 77.5 | 12.9 | - |
| Raspberry Pi 5 | ONNX Runtime FP32 | 53.6 | 1.0 | 1039.9 | - |
| Raspberry Pi 5 | PyTorch FP32 | 55.3 | 0.6 | 1638.0 | - |

## Usage with LibreYOLO

```python
from libreyolo import LibreYOLO

model = LibreYOLO("LibreYOLO9m.pt")
result = model("image.jpg", conf=0.25, iou=0.45)
```

Source: https://www.visionanalysis.org/model/yolov9m. Benchmarks produced with LibreYOLO (https://github.com/Libre-YOLO/libreyolo).