~8 min · keras-cv
from keras_hub.models import YOLOV8Detector; model = YOLOV8Detector.from_preset('yolo_v8_xs_pascalvoc') — 한 줄로 사전학습 detection. xs / s / m / l / xl size variant.
predict 후 dict 반환 — {'boxes': ..., 'classes': ..., 'confidence': ...}. Non-max-suppression (NMS) 자동. fine-tune 시 backbone freeze + head 만 학습하면 자기 dataset 의 detection 빠르게.
# Load YOLOV8 with COCO pretrained weights
detector = keras_cv.models.YOLOV8Detector.from_preset(
"yolo_v8_m_coco", # xs, s, m, l, xl sizes
)
# Run inference
predictions = detector.predict(images)
# predictions["boxes"] → bounding box coordinates
# predictions["classes"] → class IDs
# predictions["confidence"] → confidence scores아직 댓글이 없어요. 첫 댓글을 남겨보세요.