# DEIMv2-Pico

> DEIMv2-Pico: 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: 1.5M
- GFLOPs: 5.2
- Default input size: 640px
- Detection: detr · nms-free
- Architecture: transformer · DINOv3-distilled ViT backbone
- Weight license: Apache-2.0
- Paper-reported mAP@50-95: 38.5%
- 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 | 42.3 | 21.3 | 47.1 | - |
| NVIDIA Jetson Orin Nano Super 8GB | PyTorch FP32 | 42.2 | 14.1 | 71.1 | 55 |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP16 | 42.3 | 44.5 | 22.5 | - |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP32 | 42.3 | 37.7 | 26.5 | - |
| NVIDIA RTX 5070 Ti | ONNX Runtime FP32 | 42.3 | 80.2 | 12.5 | - |
| NVIDIA RTX 5070 Ti | PyTorch FP32 | 42.3 | 40.5 | 24.7 | 62 |
| NVIDIA RTX 5070 Ti | TensorRT FP32 | 42.2 | 79.9 | 12.5 | - |
| Raspberry Pi 5 | ONNX Runtime FP32 | 42.3 | 6.2 | 161.5 | - |
| Raspberry Pi 5 | PyTorch FP32 | 42.2 | 2.9 | 349.8 | - |

## Usage with LibreYOLO

```python
from libreyolo import LibreYOLO

model = LibreYOLO("LibreDEIMv2pico.pt")
result = model("image.jpg", conf=0.25, iou=0.45)
```

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