# PicoDet-S

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

- Family: PicoDet (Baidu PaddlePaddle)
- Parameters: 1.0M
- GFLOPs: 0.7
- Default input size: 320px
- Detection: anchor-free · nms
- Architecture: one-stage · ESNet backbone
- Weight license: Apache-2.0
- Paper-reported mAP@50-95: 27.1%
- Paper: https://arxiv.org/abs/2111.00902
- Original code: https://github.com/PaddlePaddle/PaddleDetection

## Benchmarks (COCO val2017)

| Hardware | Runtime | mAP@50-95 | FPS | Latency (ms) | VRAM (MB) |
|---|---|---|---|---|---|
| NVIDIA Jetson Orin Nano Super 8GB | ONNX Runtime FP32 | 29.6 | 16.4 | 61.1 | - |
| NVIDIA Jetson Orin Nano Super 8GB | PyTorch FP32 | 30.4 | 13.8 | 72.2 | 12 |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP16 | 30.4 | 39.6 | 25.3 | - |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP32 | 30.4 | 37.9 | 26.4 | - |
| NVIDIA RTX 5070 Ti | ONNX Runtime FP32 | 30.4 | 55.8 | 17.9 | - |
| NVIDIA RTX 5070 Ti | PyTorch FP32 | 30.4 | 50.0 | 20.0 | 22 |
| NVIDIA RTX 5070 Ti | TensorRT FP16 | 30.4 | 78.6 | 12.7 | - |
| NVIDIA RTX 5070 Ti | TensorRT FP32 | 30.3 | 74.0 | 13.5 | - |

## Usage with LibreYOLO

```python
from libreyolo import LibreYOLO

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

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