# DEIM-M

> DEIM-M: 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: 19.0M
- GFLOPs: 57.0
- Default input size: 640px
- Detection: detr · nms-free
- Architecture: transformer · HGNetv2 backbone
- Weight license: Apache-2.0
- Paper-reported mAP@50-95: 52.7%
- 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 | 55.4 | 6.9 | 144.6 | - |
| NVIDIA Jetson Orin Nano Super 8GB | PyTorch FP32 | 55.4 | 7.0 | 143.7 | 188 |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP16 | 55.5 | 23.5 | 42.5 | - |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP32 | 55.5 | 15.7 | 63.8 | - |
| NVIDIA RTX 5070 Ti | ONNX Runtime FP32 | 55.5 | 57.6 | 17.4 | - |
| NVIDIA RTX 5070 Ti | PyTorch FP32 | 55.5 | 28.4 | 35.3 | 189 |
| NVIDIA RTX 5070 Ti | TensorRT FP16 | 55.4 | 83.3 | 12.0 | - |
| NVIDIA RTX 5070 Ti | TensorRT FP32 | 55.5 | 85.4 | 11.7 | - |
| Raspberry Pi 5 | ONNX Runtime FP32 | 55.4 | 1.0 | 990.7 | - |
| Raspberry Pi 5 | PyTorch FP32 | 55.4 | 0.6 | 1582.8 | - |

## Usage with LibreYOLO

```python
from libreyolo import LibreYOLO

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

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