Pix2Text / tests /test_doclayout_yolo.py
fasdfsa's picture
init
2c67080
raw
history blame contribute delete
473 Bytes
# coding: utf-8
from doclayout_yolo import YOLOv10
# Load the pre-trained model
model = YOLOv10("doclayout_yolo_docstructbench_imgsz1024.pt")
img = 'docs/examples/page-authors-1.png'
# img = 'docs/examples/page.png'
# Perform prediction
det_res = model.predict(
img, # Image to predict
imgsz=1024, # Prediction image size
conf=0.2, # Confidence threshold
device="mps" # Device to use (e.g., 'cuda:0' or 'cpu')
)[0]
print(det_res.boxes)