Parameters7.2M
GFLOPs13.5
Input Size640px
Best mAP50.5%
LicenseMIT
Architecture
Type
one-stage
Backbone
GELAN
Neck
PGI
Head
Decoupled
Benchmark Results
Performance on COCO val2017 across different hardware configurations
| Hardware | Runtime | mAP@50-95 | FPS | Latency | VRAM |
|---|---|---|---|---|---|
| NVIDIA Jetson Orin Nano Super 8GB | ONNX Runtime FP32 | 50.4% | 12.5 | 79.8ms | - |
| NVIDIA Jetson Orin Nano Super 8GB | PyTorch FP32 | 50.5% | 9.9 | 101.2ms | 91 MB |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP16 | 50.4% | 33.2 | 30.1ms | - |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP32 | 50.5% | 21.6 | 46.2ms | - |
| NVIDIA RTX 5070 Ti | ONNX Runtime FP32 | 50.4% | 54.5 | 18.3ms | - |
| NVIDIA RTX 5070 Ti | PyTorch FP32 | 50.5% | 30.8 | 32.5ms | 92 MB |
| NVIDIA RTX 5070 Ti | TensorRT FP16 | 50.5% | 63.8 | 15.7ms | - |
| NVIDIA RTX 5070 Ti | TensorRT FP32 | 50.5% | 75.1 | 13.3ms | - |
Speed Breakdown(NVIDIA Jetson Orin Nano Super 8GB)
4.6ms
90.3ms
6.3ms
Preprocess
Inference
Postprocess (NMS)
Usage with LibreYOLO
from libreyolo import LibreYOLO
# Load model (auto-downloads from HuggingFace if not found locally)
model = LibreYOLO("LibreYOLO9s.pt")
# Run inference
result = model("image.jpg", conf=0.25, iou=0.45)
# Process results
print(f"Found {len(result)} objects")
print(result.boxes.xyxy) # bounding boxes (N, 4)
print(result.boxes.conf) # confidence scores (N,)
print(result.boxes.cls) # class IDs (N,)anchor-freenmsPaper: 46.8% mAP