Parameters10.0M
GFLOPs25.0
Input Size640px
Best mAP52.9%
LicenseApache-2.0
Architecture
Type
transformer
Backbone
HGNetv2
Neck
HybridEncoder
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 | 52.8% | 10.3 | 97.5ms | - |
| NVIDIA Jetson Orin Nano Super 8GB | PyTorch FP32 | 52.8% | 10.1 | 99.3ms | 139 MB |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP16 | 52.9% | 29.6 | 33.7ms | - |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP32 | 52.9% | 22.1 | 45.3ms | - |
| NVIDIA RTX 5070 Ti | ONNX Runtime FP32 | 52.8% | 66.7 | 15.0ms | - |
| NVIDIA RTX 5070 Ti | PyTorch FP32 | 52.9% | 39.8 | 25.1ms | 142 MB |
| NVIDIA RTX 5070 Ti | TensorRT FP16 | 52.9% | 92.3 | 10.8ms | - |
| NVIDIA RTX 5070 Ti | TensorRT FP32 | 52.9% | 97.1 | 10.3ms | - |
Speed Breakdown(NVIDIA Jetson Orin Nano Super 8GB)
8.8ms
87.5ms
3.0ms
Preprocess
Inference
Postprocess (NMS)
Usage with LibreYOLO
from libreyolo import LibreYOLO
# Load model (auto-downloads from HuggingFace if not found locally)
model = LibreYOLO("LibreRTDETRv4s.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: 49.7% mAP