NULLXES BLACK DEATH (UAV)
Collection
NULLXES BLACK DEATH civil UAV stack: CERBER perception models and related Hub artifacts. • 2 items • Updated
How to use MagistrTheOne/CERBER-CV-v2 with ultralytics:
from ultralytics import YOLOvv8
model = YOLOvv8.from_pretrained("MagistrTheOne/CERBER-CV-v2")
source = 'http://images.cocodataset.org/val2017/000000039769.jpg'
model.predict(source=source, save=True)Fine-tune of CERBER-CV with Seraphim drone (uav) + VisDrone keep.
| Class | P | R | mAP50 | mAP50-95 |
|---|---|---|---|---|
| all | 0.816 | 0.655 | 0.709 | 0.419 |
| human | 0.717 | 0.409 | 0.465 | 0.188 |
| vehicle | 0.811 | 0.669 | 0.735 | 0.458 |
| uav | 0.922 | 0.888 | 0.926 | 0.612 |
Train: 40 ep @640 from v1 best · ~0.43 h RTX PRO 6000.
Note: human/vehicle below v1 (prefer v1 for scene-only; v2 when UAV needed).