vishnu23/Text_Detection_and_Layout_Analysis

1import streamlit as st2import pytesseract3import cv24import pdf2image5import numpy as np6import layoutparser as lp7from PIL import Image8 9 10def text_detection(img):11  res = Image.open(img)12  res = np.array(res)13  overlay = Image.open(img)14  overlay = np.array(overlay)15  with st.spinner('Processing Text Detection'):16    boxes = pytesseract.image_to_data(res)17    for i,b in enumerate(boxes.splitlines()): 18      if i!=0:19        b = b.split()20        if len(b)==12:21          x,y,w,h = int(b[6]),int(b[7]),int(b[8]),int(b[9])22          cv2.rectangle(overlay,(x-5,y-5),(w+x+5,h+y+5),(185,237,221),-1) 23          #res = cv2.putText(res,b[11],(x,y),cv2.FONT_HERSHEY_COMPLEX,1,(185,237,221),2)24  new = cv2.addWeighted(overlay, 0.4, res, 1 - 0.4, 0)25  st.subheader("Text Detection")26  st.image(new)27 28def parserIMG(path):29  res = Image.open(path)30  with st.spinner('Processing Layout Analysis'):31    model = lp.Detectron2LayoutModel('lp://PubLayNet/faster_rcnn_R_50_FPN_3x/config',32                                  extra_config=["MODEL.ROI_HEADS.SCORE_THRESH_TEST", 0.5],33                                  label_map={0: "Text", 1: "Title", 2: "List", 3:"Table", 4:"Figure"})34    layout_result = model.detect(res)35    res=lp.draw_box(res, layout_result,  box_width=5, box_alpha=0.2, show_element_type=True)36  st.subheader("Layout Analysis")37  st.image(res)38 39def parserPDF(path):40  img = np.asarray(pdf2image.convert_from_bytes(path.read()))41  st.subheader("Layout Analysis")42 43  with st.spinner('Processing Layout Analysis'):44    model = lp.Detectron2LayoutModel('lp://PubLayNet/faster_rcnn_R_50_FPN_3x/config',45                                  extra_config=["MODEL.ROI_HEADS.SCORE_THRESH_TEST", 0.5],46                                  label_map={0: "Text", 1: "Title", 2: "List", 3:"Table", 4:"Figure"})47    for i in img:48      layout_result = model.detect(i)49      i=lp.draw_box(i, layout_result,  box_width=5, box_alpha=0.2, show_element_type=True)50      st.image(i)51 52#title 53st.title('Text Detection & Layout Analysis')54#uploading file55file = st.file_uploader(label='Upload your document here',type=['png','jpg','jpeg','pdf'])56button = st.button('Confirm')57if button and file is not None:58    if file.type=="image/png" or file.type=="image/jpg" or file.type=="image/jpeg":59      parserIMG(file)60      text_detection(file)61    elif file.type == "application/pdf":62      parserPDF(file)