# YOLOv5lu

> YOLOv5lu: one-stage object detector from the yolov5u family. Planned for LibreYOLO (not yet packaged).

- Family: yolov5u
- Parameters: 53.2M
- GFLOPs: 135.0
- Default input size: 640px
- Detection: anchor-free · nms
- Architecture: one-stage · CSPDarknet backbone
- Weight license: non-permissive
- Paper-reported mAP@50-95: 52.2%

## Benchmarks (COCO val2017)

| Hardware | Runtime | mAP@50-95 | FPS | Latency (ms) | VRAM (MB) |
|---|---|---|---|---|---|
| NVIDIA Jetson Orin Nano Super 8GB | ONNX Runtime FP32 | 51.5 | 0.8 | 1247.2 | - |
| NVIDIA Jetson Orin Nano Super 8GB | PyTorch FP32 | 51.5 | 8.6 | 116.9 | 264 |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP16 | 51.5 | 20.6 | 48.5 | 14 |
| Raspberry Pi 5 | ncnn FP32 | 51.5 | 1.6 | 608.2 | - |
| Raspberry Pi 5 | ONNX Runtime FP32 | 51.5 | 0.6 | 1625.4 | - |
| Raspberry Pi 5 | PyTorch FP32 | 51.5 | 0.6 | 1587.0 | - |


Source: https://www.visionanalysis.org/model/yolov5lu. Benchmarks produced with LibreYOLO (https://github.com/Libre-YOLO/libreyolo).