Parameters31.0M
GFLOPs92.0
Input Size640px
Best mAP53.2%
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 | 53.2% | 7.6 | 131.5ms | - |
| NVIDIA Jetson Orin Nano Super 8GB | PyTorch FP32 | 53.2% | 8.1 | 123.1ms | 199 MB |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP16 | 53.2% | 32.7 | 30.6ms | - |
| NVIDIA Jetson Orin Nano Super 8GB | TensorRT FP32 | 53.2% | 17.3 | 58.0ms | - |
| NVIDIA RTX 5070 Ti | ONNX Runtime FP32 | 53.2% | 81.1 | 12.3ms | - |
| NVIDIA RTX 5070 Ti | PyTorch FP32 | 53.2% | 40.4 | 24.8ms | 213 MB |
| NVIDIA RTX 5070 Ti | TensorRT FP16 | 53.2% | 132.9 | 7.5ms | - |
| NVIDIA RTX 5070 Ti | TensorRT FP32 | 53.2% | 102.5 | 9.8ms | - |
Speed Breakdown(NVIDIA Jetson Orin Nano Super 8GB)
4.5ms
115.9ms
2.7ms
Preprocess
Inference
Postprocess (NMS)
Usage with LibreYOLO
from libreyolo import LibreYOLO
# Load model (auto-downloads from HuggingFace if not found locally)
model = LibreYOLO("LibreRTDETRv2r34.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: 49.9% mAP