import os from PIL import Image, ImageOps, ImageChops import io import fitz # PyMuPDF from docx import Document from rembg import remove import gradio as gr def trim_whitespace(image): gray_image = ImageOps.grayscale(image) inverted_image = ImageChops.invert(gray_image) bbox = inverted_image.getbbox() trimmed_image = image.crop(bbox) return trimmed_image def convert_pdf_to_images(pdf_path, zoom=2): pdf_document = fitz.open(pdf_path) images = [] for page_num in range(len(pdf_document)): page = pdf_document.load_page(page_num) matrix = fitz.Matrix(zoom, zoom) pix = page.get_pixmap(matrix=matrix) image = Image.frombytes("RGB", [pix.width, pix.height], pix.samples) trimmed_image = trim_whitespace(image) images.append(trimmed_image) return images def convert_docx_to_images(docx_path): document = Document(docx_path) images = [] for image_shape in document.inline_shapes: image_stream = image_shape.image.blob image = Image.open(io.BytesIO(image_stream)) images.append(image) return images def remove_background_from_image(image): return remove(image) def process_file(input_file): file_extension = os.path.splitext(input_file.name)[1].lower() if file_extension in ['.png', '.jpeg', '.jpg', '.bmp', '.gif']: image = Image.open(input_file) image = image.convert('RGB') output_image = remove_background_from_image(image) return output_image elif file_extension == '.pdf': images = convert_pdf_to_images(input_file.name) return [remove_background_from_image(image) for image in images] elif file_extension in ['.docx', '.doc']: images = convert_docx_to_images(input_file.name) return [remove_background_from_image(image) for image in images] else: return "File format not supported." def gradio_interface(input_file): return process_file(input_file) iface = gr.Interface( fn=gradio_interface, inputs=gr.File(label="Upload Word, PDF, or Image"), outputs=gr.outputs.Image(type="file", label="Processed Image(s)"), title="Document to Image Converter with Background Removal" ) if __name__ == "__main__": iface.launch()