| import copy | |
| from src.utils import run_example, clean_text, draw_ocr_bboxes | |
| def ocr_task(image): | |
| """ | |
| Perform OCR (Optical Character Recognition) on the given image. | |
| Args: | |
| image (PIL.Image.Image): The input image to perform OCR on. | |
| Returns: | |
| tuple: A tuple containing the output image with OCR bounding boxes drawn and the cleaned OCR text. | |
| """ | |
| # Task prompts | |
| ocr_prompt = "<OCR>" | |
| ocr_with_region_prompt = "<OCR_WITH_REGION>" | |
| # Get OCR text | |
| ocr_results = run_example(ocr_prompt, image) | |
| cleaned_text = clean_text(ocr_results["<OCR>"]) | |
| # Get OCR with region | |
| ocr_with_region_results = run_example(ocr_with_region_prompt, image) | |
| output_image = copy.deepcopy(image) | |
| output_image = draw_ocr_bboxes( | |
| output_image, ocr_with_region_results["<OCR_WITH_REGION>"] | |
| ) | |
| # Return the output image and cleaned OCR text | |
| return output_image, cleaned_text | |