Back to Leaderboard
Parameters
43.7M
FLOPs
165.2G
Input Size
640px
License
MIT
Architecture
Type
one-stage
Backbone
CSPDarknet
Neck
PAFPN
Head
Decoupled
Benchmark Results
Performance on COCO val2017 across different hardware configurations
| Hardware | mAP@50-95 | FPS | Latency | VRAM |
|---|---|---|---|---|
| NVIDIA A100 (TensorRT FP16) | 52.9% | 82.5 | 12.1ms | 1862 MB |
| NVIDIA T4 (TensorRT FP16) | 53.0% | 32.0 | 31.2ms | 1883 MB |
| CPU (ONNX Runtime) | 53.0% | 3.7 | 272.0ms | 1911 MB |
Speed Breakdown (A100 TensorRT)
End-to-end latency breakdown showing preprocessing, inference, and postprocessing times
1.4ms
8.1ms
2.7ms
Preprocess
Inference
Postprocess (NMS)
Usage with LibreYOLO
from libreyolo import YOLO
# Load model
model = YOLO.from_pretrained("https://huggingface.co/Libre-YOLO/yolov8l")
# Run inference
results = model.predict("image.jpg")
# Process results
for box in results.boxes:
print(f"Class: {box.cls}, Confidence: {box.conf:.2f}")production-readyhigh-accuracy