p7ivanov/rose-disease-detection

Rose Disease Detection — YOLOv11n

AI model for detecting diseases on Rosa damascena (Bulgarian oil-bearing rose).

Status: Early test version (v1)
This model is still in training and may produce errors.
We are continuously improving it with new data from real parcels.

What it detects

Class	Bulgarian	Notes
Black Spot	Черни петна	Marssonina rosae
Downy Mildew	Мана	Peronospora sparsa
Powdery Mildew	Брашнеста мана	Podosphaera pannosa
Normal	Здраво растение	No disease detected

Coming in future versions: pests (aphids, spider mites), weeds.

Results

• mAP50 = 0.921
• Precision = 0.925
• Recall = 0.884
• Architecture: YOLOv11n
• Training: 50 epochs, 640px, batch 16

How to use

from ultralytics import YOLO

model = YOLO("piivanov77/rose-disease-detection/best_v1.pt")
results = model.predict("your_image.jpg", conf=0.25)

for r in results:
    for box in r.boxes:
        print(model.names[int(box.cls[0])], f"{float(box.conf[0]):.0%}")

Data Sources

Rose Disease Prediction Dataset

• Author: vinodk
• Source: https://universe.roboflow.com/vinodk-cb0f7/rose-disease-prediction-yolov5
• License: CC BY 4.0
• Images: 3702, Classes: Black Spot, Downy Mildew, Normal, Powdery Mildew

License

CC BY 4.0 — free to use with attribution.

Citation

Rose Disease Detection YOLOv11 — Rosa damascena
Trained by Pavel Ivanov, 2026
Dataset: vinodk @ Roboflow Universe (CC BY 4.0)
https://universe.roboflow.com/vinodk-cb0f7/rose-disease-prediction-yolov5