Back to Leaderboard
Parameters
25.9M
FLOPs
78.9G
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) | 50.2% | 112.8 | 8.9ms | 1077 MB |
| NVIDIA T4 (TensorRT FP16) | 50.2% | 41.6 | 24.1ms | 1185 MB |
| CPU (ONNX Runtime) | 50.2% | 5.3 | 188.8ms | 1086 MB |
Speed Breakdown (A100 TensorRT)
End-to-end latency breakdown showing preprocessing, inference, and postprocessing times
1.3ms
5.1ms
2.5ms
Preprocess
Inference
Postprocess (NMS)
Usage with LibreYOLO
from libreyolo import YOLO
# Load model
model = YOLO.from_pretrained("https://huggingface.co/Libre-YOLO/yolov8m")
# Run inference
results = model.predict("image.jpg")
# Process results
for box in results.boxes:
print(f"Class: {box.cls}, Confidence: {box.conf:.2f}")production-readybalanced