Parameters33.9M
GFLOPs0.0
Input Size704px
Best mAP58.7%
LicenseApache-2.0
Architecture
Type
transformer
Backbone
DINOv2
Neck
Agg2Former
Head
DETR
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 | 58.6% | 3.4 | 295.5ms | - |
| NVIDIA Jetson Orin Nano Super 8GB | PyTorch FP32 | 58.5% | 4.0 | 248.5ms | 222 MB |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP16 | 58.7% | 8.4 | 119.3ms | - |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP32 | 58.6% | 7.8 | 127.8ms | - |
| NVIDIA RTX 5070 Ti | ONNX Runtime FP32 | 58.6% | 43.8 | 22.8ms | - |
| NVIDIA RTX 5070 Ti | PyTorch FP32 | 58.5% | 25.1 | 39.9ms | 223 MB |
| NVIDIA RTX 5070 Ti | TensorRT FP16 | 58.6% | 47.6 | 21.0ms | - |
| NVIDIA RTX 5070 Ti | TensorRT FP32 | 58.6% | 45.1 | 22.2ms | - |
Speed Breakdown(NVIDIA Jetson Orin Nano Super 8GB)
21.4ms
222.1ms
5.0ms
Preprocess
Inference
Postprocess (NMS)
Usage with LibreYOLO
from libreyolo import LibreYOLO
# Load model (auto-downloads from HuggingFace if not found locally)
model = LibreYOLO("LibreRFDETRl.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,)detrnms-freePaper: 56.5% mAP