Parameters42.0M
GFLOPs136.0
Input Size640px
Best mAP55.7%
LicenseApache-2.0
Architecture
Type
transformer
Backbone
PResNet
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 | 55.7% | 4.4 | 227.4ms | - |
| NVIDIA Jetson Orin Nano Super 8GB | PyTorch FP32 | 55.7% | 4.3 | 233.4ms | 323 MB |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP16 | 55.7% | 24.2 | 41.3ms | - |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP32 | 55.7% | 12.7 | 78.6ms | - |
| NVIDIA RTX 5070 Ti | ONNX Runtime FP32 | 55.6% | 66.0 | 15.2ms | - |
| NVIDIA RTX 5070 Ti | PyTorch FP32 | 55.7% | 32.1 | 31.1ms | 327 MB |
| NVIDIA RTX 5070 Ti | TensorRT FP16 | 55.7% | 103.6 | 9.7ms | - |
| NVIDIA RTX 5070 Ti | TensorRT FP32 | 55.7% | 88.5 | 11.3ms | - |
Speed Breakdown(NVIDIA Jetson Orin Nano Super 8GB)
4.5ms
226.2ms
2.8ms
Preprocess
Inference
Postprocess (NMS)
Usage with LibreYOLO
from libreyolo import LibreYOLO
# Load model (auto-downloads from HuggingFace if not found locally)
model = LibreYOLO("LibreRTDETRv2r50.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: 53.4% mAP