Back to Leaderboard
Parameters
6.2M
FLOPs
13.7G
Input Size
640px
License
GPL-3.0
Architecture
Type
one-stage
Backbone
E-ELAN
Neck
SPPCSPC
Head
RepConv
Benchmark Results
Performance on COCO val2017 across different hardware configurations
| Hardware | mAP@50-95 | FPS | Latency | VRAM |
|---|---|---|---|---|
| NVIDIA A100 (TensorRT FP16) | 37.4% | 184.8 | 5.4ms | 444 MB |
| NVIDIA T4 (TensorRT FP16) | 37.3% | 74.4 | 13.4ms | 301 MB |
| CPU (ONNX Runtime) | 37.4% | 9.0 | 111.7ms | 277 MB |
Speed Breakdown (A100 TensorRT)
End-to-end latency breakdown showing preprocessing, inference, and postprocessing times
1.3ms
1.8ms
2.3ms
Preprocess
Inference
Postprocess (NMS)
Usage with LibreYOLO
from libreyolo import YOLO
# Load model
model = YOLO.from_pretrained("https://huggingface.co/Libre-YOLO/yolov7-tiny")
# Run inference
results = model.predict("image.jpg")
# Process results
for box in results.boxes:
print(f"Class: {box.cls}, Confidence: {box.conf:.2f}")edge-friendlyefficient