Back to Leaderboard

YOLOv9-E

yolov9

one-stage detector with GELAN backbone

Parameters

58.1M

FLOPs

192.5G

Input Size

640px

License

MIT

Architecture

Type

one-stage

Backbone

GELAN

Neck

PGI

Head

Decoupled

Benchmark Results
Performance on COCO val2017 across different hardware configurations
HardwaremAP@50-95FPSLatencyVRAM
NVIDIA A100 (TensorRT FP16)55.5%70.714.1ms2579 MB
NVIDIA T4 (TensorRT FP16)55.6%25.539.2ms2610 MB
CPU (ONNX Runtime)55.7%3.0331.9ms2549 MB
Speed Breakdown (A100 TensorRT)
End-to-end latency breakdown showing preprocessing, inference, and postprocessing times
1.0ms
10.7ms
2.4ms
Preprocess
Inference
Postprocess (NMS)
Usage with LibreYOLO
from libreyolo import YOLO

# Load model
model = YOLO.from_pretrained("https://huggingface.co/Libre-YOLO/yolov9e")

# Run inference
results = model.predict("image.jpg")

# Process results
for box in results.boxes:
    print(f"Class: {box.cls}, Confidence: {box.conf:.2f}")
programmable-gradienthighest-accuracy
Notes

Highest accuracy YOLOv9 variant

Related Models (yolov9)