All models were trained on 2 * T4
| Model | Precision | Recall |
|---|---|---|
| YOLOv8n | 92.5 | 81.4 |
| YOLOv9t | 92.2 | 80.8 |
| YOLOv10n | 93.3 | 78.9 |
| YOLOv11n | 92.8 | 80.4 |
| YOLOv12n | 92.4 | 79.2 |
| YOLOv26n | ? | ? |
| Model | Precision | Recall |
|---|---|---|
| YOLOv8s | 94.3 | 85.7 |
| YOLOv9s | 93.7 | 84.3 |
| YOLOv10s | 93.1 | 84.1 |
| YOLOv11s | 93 | 85.2 |
| YOLOv12s | 94 | 82.9 |
| YOLOv26s | 95.4 | 87.7 |
| Model | Precision | Recall |
|---|---|---|
| YOLOv8m | 93.7 | 87.4 |
| YOLOv11m | 94.5 | 87.2 |
| YOLOv12m | ? | ? |
| YOLOv26m | ? | ? |
Epoch=100
batch=16
optimizer=SGD
lr=0.01
momentum=0.937
dataset: Volleyvision
train set: ~17000 images
validation set: ~5000 images
- ultralytics>=8.3
- numpy
- torch
- collections
