# DEIM-N

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

- Family: DEIM (Intellindust AI Lab)
- Parameters: 4.0M
- GFLOPs: 7.0
- Default input size: 640px
- Detection: detr · nms-free
- Architecture: transformer · HGNetv2 backbone
- Weight license: Apache-2.0
- Paper-reported mAP@50-95: 43%
- Paper: https://arxiv.org/abs/2412.04234
- Original code: https://github.com/Intellindust-AI-Lab/DEIM

## Benchmarks (COCO val2017)

| Hardware | Runtime | mAP@50-95 | FPS | Latency (ms) | VRAM (MB) |
|---|---|---|---|---|---|
| NVIDIA Jetson Orin Nano Super 8GB | ONNX Runtime FP32 | 46.8 | 17.5 | 57.2 | - |
| NVIDIA Jetson Orin Nano Super 8GB | PyTorch FP32 | 46.8 | 11.3 | 88.5 | 64 |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP16 | 46.8 | 41.8 | 23.9 | - |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP32 | 46.8 | 33.4 | 30.0 | - |
| NVIDIA RTX 5070 Ti | ONNX Runtime FP32 | 46.8 | 75.9 | 13.2 | - |
| NVIDIA RTX 5070 Ti | PyTorch FP32 | 46.8 | 37.0 | 27.0 | 70 |
| NVIDIA RTX 5070 Ti | TensorRT FP16 | 46.6 | 81.3 | 12.3 | - |
| NVIDIA RTX 5070 Ti | TensorRT FP32 | 46.8 | 67.5 | 14.8 | - |
| Raspberry Pi 5 | ONNX Runtime FP32 | 46.8 | 4.3 | 229.7 | - |
| Raspberry Pi 5 | PyTorch FP32 | 46.8 | 2.3 | 431.0 | - |

## Usage with LibreYOLO

```python
from libreyolo import LibreYOLO

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

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