Label_studio_to_layoutLMV3.py

import json

def convert_bounding_box(x, y, width, height):
	"""Converts the given bounding box coordinates to the YOLO format.

	Args:
	x: The x-coordinate of the top-left corner of the bounding box.
	y: The y-coordinate of the top-left corner of the bounding box.
	width: The width of the bounding box.
	height: The height of the bounding box.

	Returns:
	A tuple of four coordinates (x1, y1, x2, y2) in the YOLO format.
	"""

	x1 = x
	y1 = y
	x2 = x + width
	y2 = y + height

	return [x1, y1, x2, y2]



####################################### Loading json data ###################################
with open("D:/Projects/AI_Projects/NLP/Document_AI/LayoutLM_Models/Training_json.json") as f:
    data = json.load(f)


output = []

for annoatated_image in data:
	data = {}
	annotation = []
	ann_list = []

	if len(annoatated_image) < 8:
		continue

	for k, v in annoatated_image.items():
		if k == 'ocr':
			v = v.split('8080/')[-1]
			print(f'filename: {v}')

			data["file_name"] = f"D:/Projects/AI_Projects/NLP/Document_AI/LayoutLM_Models/image/Training_Images/{v}"
			output.append(data)


		if k == 'bbox':
			width = v[0]['original_width']
			height = v[0]['original_height']

			data["height"] = height
			data["width"] = width



	for bb, text, label in zip(annoatated_image['bbox'], annoatated_image['transcription'],   annoatated_image['label']):
		ann_dict = {}

		print('text :', text)

		ann_dict["box"] = convert_bounding_box(bb['x'], bb['y'], bb['width'], bb['height'])
		ann_dict["text"] = text
		ann_dict["label"] = label['labels'][-1]
		ann_list.append(ann_dict)
	data["annotations"] = ann_list

print(output)
with open("Training_layoutLMV3.json", "w") as f:
  json.dump(output, f, indent=4)