# DEIMv2-X

> DEIMv2-X: transformer object detector from the DEIMv2 family. Available in LibreYOLO, the MIT-licensed open-source library (free for commercial use).

- Family: DEIMv2 (Intellindust AI Lab)
- Parameters: 50.3M
- GFLOPs: 151.6
- Default input size: 640px
- Detection: detr · nms-free
- Architecture: transformer · DINOv3-distilled ViT backbone
- Weight license: Apache-2.0
- Paper-reported mAP@50-95: 57.8%
- Paper: https://arxiv.org/abs/2509.20787
- Original code: https://github.com/Intellindust-AI-Lab/DEIMv2

## Benchmarks (COCO val2017)

| Hardware | Runtime | mAP@50-95 | FPS | Latency (ms) | VRAM (MB) |
|---|---|---|---|---|---|
| NVIDIA Jetson Orin Nano Super 8GB | ONNX Runtime FP32 | 61.3 | 2.3 | 444.9 | - |
| NVIDIA Jetson Orin Nano Super 8GB | PyTorch FP32 | 61.3 | 2.7 | 370.1 | 294 |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP16 | 60.8 | 5.4 | 184.5 | - |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP32 | 61.3 | 5.2 | 194.2 | - |
| NVIDIA RTX 5070 Ti | ONNX Runtime FP32 | 61.4 | 27.1 | 36.9 | - |
| NVIDIA RTX 5070 Ti | PyTorch FP32 | 61.3 | 17.4 | 57.6 | 292 |
| NVIDIA RTX 5070 Ti | TensorRT FP16 | 60.8 | 37.9 | 26.4 | - |
| NVIDIA RTX 5070 Ti | TensorRT FP32 | 61.3 | 38.1 | 26.2 | - |
| Raspberry Pi 5 | ONNX Runtime FP32 | 61.3 | 0.3 | 3453.0 | - |
| Raspberry Pi 5 | PyTorch FP32 | 61.3 | 0.3 | 3711.8 | - |

## Usage with LibreYOLO

```python
from libreyolo import LibreYOLO

model = LibreYOLO("LibreDEIMv2x.pt")
result = model("image.jpg", conf=0.25, iou=0.45)
```

Source: https://www.visionanalysis.org/model/deimv2-x. Benchmarks produced with LibreYOLO (https://github.com/Libre-YOLO/libreyolo).