推理源码
import cv2
from doclayout_yolo import YOLOv10
# Load the pre-trained model
model = YOLOv10("doclayout_yolo_docstructbench_imgsz1024.pt")
# Perform prediction
det_res = model.predict(
"1.JPG", # Image to predict
imgsz=1024, # Prediction image size
conf=0.2, # Confidence threshold
device="cpu" # Device to use (e.g., 'cuda:0' or 'cpu')
)
# Annotate and save the result
annotated_frame = det_res[0].plot(pil=True, line_width=5, font_size=20)
cv2.imwrite("result.jpg", annotated_frame)