34 lines
934 B
Python
34 lines
934 B
Python
"""Sonuclari filtrele, YOLO/COCO/CSV/JSON olarak kaydet."""
|
|
|
|
from evren_sdk import EvrenClient
|
|
|
|
client = EvrenClient(api_key="evren_xxxxx")
|
|
|
|
result = client.predict("kullanici/model", "sahne.jpg", confidence=0.1)
|
|
print(f"Ham: {result.count} tespit\n")
|
|
|
|
# filtrele
|
|
yuksek = result.filter(min_confidence=0.7)
|
|
print(f"conf >= 0.7 : {yuksek.count}")
|
|
|
|
sadece_araba = result.filter(classes=["araba"])
|
|
print(f"araba : {sadece_araba.count}")
|
|
|
|
combo = result.filter(min_confidence=0.5, max_confidence=0.9, classes=["insan"])
|
|
print(f"insan 0.5-0.9: {combo.count}")
|
|
|
|
print("\n-- YOLO --")
|
|
print(result.to_yolo()[:200])
|
|
|
|
print("\n-- COCO --")
|
|
for ann in result.to_coco()[:3]:
|
|
print(f" cat={ann['category_id']} bbox={ann['bbox']} area={ann['area']:.0f}")
|
|
|
|
print("\n-- CSV --")
|
|
print(result.to_csv()[:300])
|
|
|
|
# uzantiya gore format secer
|
|
result.save("output/sonuc.json")
|
|
result.save("output/sonuc.csv")
|
|
result.save("output/labels.txt")
|