Use streamlit_option_menu

Home.py

@@ -1,19 +1,17 @@
 import streamlit as st
+from multipage import MultiPage
+from app_pages import home, about, ocr_comparator
 
-st.set_page_config(page_title='OCR Comparator', layout ="wide")
-st.image('ocr.png')
+app = MultiPage()
+st.set_page_config(
+    page_title='OCR Comparator', layout ="wide",
+    initial_sidebar_state="expanded",
+)
 
-st.write("")
+# Add all your application here
+app.add_page("Home", "house", home.app)
+app.add_page("About", "info-circle", about.app)
+app.add_page("App", "cast", ocr_comparator.app)
 
-st.markdown('''#### OCR, or Optical Character Recognition, is a computer vision task, \
-which includes the detection of text areas, and the recognition of characters.''')
-st.write("")
-st.write("")
-
-st.markdown("#####  This app allows you to compare, from a given image, the results of different solutions:")
-st.markdown("##### *EasyOcr, PaddleOCR, MMOCR, Tesseract*")
-st.write("")
-st.write("")
-st.markdown("👈 Select the **About** page from the sidebar for information on how the app works")
-
-st.markdown("👈 or directly select the **App** page")
+# The main app
+app.run()

app_pages/about.py

@@ -0,0 +1,37 @@
+import streamlit as st
+
+def app():
+    st.title("OCR solutions comparator")
+
+    st.write("")
+    st.write("")
+    st.write("")
+
+    st.markdown("#####  This app allows you to compare, from a given picture, the results of different solutions:")
+    st.markdown("##### *EasyOcr, PaddleOCR, MMOCR, Tesseract*")
+    st.write("")
+    st.write("")
+
+    st.markdown(''' The 1st step is to choose the language for the text recognition (not all solutions \
+    support the same languages), and then choose the picture to consider. It is possible to upload a file, \
+    to take a picture, or to use a demo file. \
+    It is then possible to change the default values for the text area detection process, \
+    before launching the detection task for each solution.''')
+    st.write("")
+
+    st.markdown(''' The different results are then presented. The 2nd step is to choose one of these \
+    detection results, in order to carry out the text recognition process there. It is also possible to change \
+    the default settings for each solution.''')
+    st.write("")
+
+    st.markdown("###### The recognition results appear in 2 formats:")
+    st.markdown(''' - a visual format resumes the initial image, replacing the detected areas with \
+    the recognized text. The background is + or - strongly colored in green according to the \
+    confidence level of the recognition.
+        A slider allows you to change the font size, another \
+    allows you to modify the confidence threshold above which the text color changes: if it is at \
+    70% for example, then all the texts with a confidence threshold higher or equal to 70 will appear \
+    in white, in black otherwise.''')
+
+    st.markdown(" - a detailed format presents the results in a table, for each text box detected. \
+    It is possible to download this results in a local csv file.")

app_pages/home.py

@@ -0,0 +1,19 @@
+import streamlit as st
+
+def app():
+    st.image('ocr.png')
+
+    st.write("")
+
+    st.markdown('''#### OCR, or Optical Character Recognition, is a computer vision task, \
+    which includes the detection of text areas, and the recognition of characters.''')
+    st.write("")
+    st.write("")
+
+    st.markdown("#####  This app allows you to compare, from a given image, the results of different solutions:")
+    st.markdown("##### *EasyOcr, PaddleOCR, MMOCR, Tesseract*")
+    st.write("")
+    st.write("")
+    st.markdown("👈 Select the **About** page from the sidebar for information on how the app works")
+
+    st.markdown("👈 or directly select the **App** page")

pages/App.py → app_pages/ocr_comparator.py

@@ -929,491 +929,482 @@ def raz():
 ###################################################################################################
 ##   MAIN
 ###################################################################################################
+def app():
+    ##----------- Initializations ---------------------------------------------------------------------
+    #print("PID : ", os.getpid())
+
+    st.title("OCR solutions comparator")
+    st.markdown("##### *EasyOCR, PPOCR, MMOCR, Tesseract*")
+    #st.markdown("#### PID : " + str(os.getpid()))
+
+    # Initializations
+    with st.spinner("Initializations in progress ..."):
+        reader_type_list, reader_type_dict, list_dict_lang, \
+        cols_size, dict_back_colors, fig_colorscale = initializations()
+        img_demo_1, img_demo_2 = get_demo()
+
+    ##----------- Choose language & image -------------------------------------------------------------
+    st.markdown("#### Choose languages for the text recognition:")
+    lang_col = st.columns(4)
+    easyocr_key_lang = lang_col[0].selectbox(reader_type_list[0]+" :", list_dict_lang[0].keys(), 26)
+    easyocr_lang = list_dict_lang[0][easyocr_key_lang]
+    ppocr_key_lang = lang_col[1].selectbox(reader_type_list[1]+" :", list_dict_lang[1].keys(), 22)
+    ppocr_lang = list_dict_lang[1][ppocr_key_lang]
+    mmocr_key_lang = lang_col[2].selectbox(reader_type_list[2]+" :", list_dict_lang[2].keys(), 0)
+    mmocr_lang = list_dict_lang[2][mmocr_key_lang]
+    tesserocr_key_lang = lang_col[3].selectbox(reader_type_list[3]+" :", list_dict_lang[3].keys(), 35)
+    tesserocr_lang = list_dict_lang[3][tesserocr_key_lang]
+
+    st.markdown("#### Choose picture:")
+    cols_pict = st.columns([1, 2])
+    img_typ = cols_pict[0].radio("", ['Upload file', 'Take a picture', 'Use a demo file'], \
+                                index=0, on_change=raz)
+
+    if img_typ == 'Upload file':
+        image_file = cols_pict[1].file_uploader("Upload a file:", type=["jpg","jpeg"], on_change=raz)
+    if img_typ == 'Take a picture':
+        image_file = cols_pict[1].camera_input("Take a picture:", on_change=raz)
+    if img_typ == 'Use a demo file':
+        with st.expander('Choose a demo file:', expanded=True):
+            demo_used = st.radio('', ['File 1', 'File 2'], index=0, \
+                                horizontal=True, on_change=raz)
+            cols_demo = st.columns([1, 2])
+            cols_demo[0].markdown('###### File 1')
+            cols_demo[0].image(img_demo_1, width=150)
+            cols_demo[1].markdown('###### File 2')
+            cols_demo[1].image(img_demo_2, width=300)
+            if demo_used == 'File 1':
+                image_file = 'img_demo_1.jpg'
+            else:
+                image_file = 'img_demo_2.jpg'
 
-##----------- Initializations ---------------------------------------------------------------------
-#print("PID : ", os.getpid())
-st.set_page_config(page_title='OCR Comparator', layout ="wide")
-st.markdown("""
-    <style>
-        section[data-testid="stSidebar"] {
-            width: 5rem;
-            }
-    </style>
-""",unsafe_allow_html=True)
-
-
-st.title("OCR solutions comparator")
-st.markdown("##### *EasyOCR, PPOCR, MMOCR, Tesseract*")
-#st.markdown("#### PID : " + str(os.getpid()))
-
-# Initializations
-with st.spinner("Initializations in progress ..."):
-    reader_type_list, reader_type_dict, list_dict_lang, \
-    cols_size, dict_back_colors, fig_colorscale = initializations()
-    img_demo_1, img_demo_2 = get_demo()
-
-##----------- Choose language & image -------------------------------------------------------------
-st.markdown("#### Choose languages for the text recognition:")
-lang_col = st.columns(4)
-easyocr_key_lang = lang_col[0].selectbox(reader_type_list[0]+" :", list_dict_lang[0].keys(), 26)
-easyocr_lang = list_dict_lang[0][easyocr_key_lang]
-ppocr_key_lang = lang_col[1].selectbox(reader_type_list[1]+" :", list_dict_lang[1].keys(), 22)
-ppocr_lang = list_dict_lang[1][ppocr_key_lang]
-mmocr_key_lang = lang_col[2].selectbox(reader_type_list[2]+" :", list_dict_lang[2].keys(), 0)
-mmocr_lang = list_dict_lang[2][mmocr_key_lang]
-tesserocr_key_lang = lang_col[3].selectbox(reader_type_list[3]+" :", list_dict_lang[3].keys(), 35)
-tesserocr_lang = list_dict_lang[3][tesserocr_key_lang]
-
-st.markdown("#### Choose picture:")
-cols_pict = st.columns([1, 2])
-img_typ = cols_pict[0].radio("", ['Upload file', 'Take a picture', 'Use a demo file'], \
-                             index=0, on_change=raz)
-
-if img_typ == 'Upload file':
-    image_file = cols_pict[1].file_uploader("Upload a file:", type=["jpg","jpeg"], on_change=raz)
-if img_typ == 'Take a picture':
-    image_file = cols_pict[1].camera_input("Take a picture:", on_change=raz)
-if img_typ == 'Use a demo file':
-    with st.expander('Choose a demo file:', expanded=True):
-        demo_used = st.radio('', ['File 1', 'File 2'], index=0, \
-                             horizontal=True, on_change=raz)
-        cols_demo = st.columns([1, 2])
-        cols_demo[0].markdown('###### File 1')
-        cols_demo[0].image(img_demo_1, width=150)
-        cols_demo[1].markdown('###### File 2')
-        cols_demo[1].image(img_demo_2, width=300)
-        if demo_used == 'File 1':
-            image_file = 'img_demo_1.jpg'
-        else:
-            image_file = 'img_demo_2.jpg'
-
-##----------- Process input image -----------------------------------------------------------------
-if image_file is not None:
-    image_path, image_orig, image_cv2 = load_image(image_file)
-    list_images = [image_orig, image_cv2]
-
-##----------- Form with original image & hyperparameters for detectors ----------------------------
-    with st.form("form1"):
-        col1, col2 = st.columns(2, ) #gap="medium")
-        col1.markdown("##### Original image")
-        col1.image(list_images[0], width=500, use_column_width=True)
-        col2.markdown("##### Hyperparameters values for detection")
-
-        with col2.expander("Choose detection hyperparameters for " + reader_type_list[0], \
-                           expanded=False):
-            t0_min_size = st.slider("min_size", 1, 20, 10, step=1, \
-                    help="min_size (int, default = 10) - Filter text box smaller than \
-                          minimum value in pixel")
-            t0_text_threshold = st.slider("text_threshold", 0.1, 1., 0.7, step=0.1, \
-                    help="text_threshold (float, default = 0.7) - Text confidence threshold")
-            t0_low_text = st.slider("low_text", 0.1, 1., 0.4, step=0.1, \
-                    help="low_text (float, default = 0.4) - Text low-bound score")
-            t0_link_threshold = st.slider("link_threshold", 0.1, 1., 0.4, step=0.1, \
-                    help="link_threshold (float, default = 0.4) - Link confidence threshold")
-            t0_canvas_size = st.slider("canvas_size", 2000, 5000, 2560, step=10, \
-                    help='''canvas_size (int, default = 2560) \n
-Maximum e size. Image bigger than this value will be resized down''')
-            t0_mag_ratio = st.slider("mag_ratio", 0.1, 5., 1., step=0.1, \
-                    help="mag_ratio (float, default = 1) - Image magnification ratio")
-            t0_slope_ths = st.slider("slope_ths", 0.01, 1., 0.1, step=0.01, \
-                    help='''slope_ths (float, default = 0.1) - Maximum slope \
-                            (delta y/delta x) to considered merging. \n
-Low valuans tiled boxes will not be merged.''')
-            t0_ycenter_ths = st.slider("ycenter_ths", 0.1, 1., 0.5, step=0.1, \
-                    help='''ycenter_ths (float, default = 0.5) - Maximum shift in y direction. \n
-Boxes wiifferent level should not be merged.''')
-            t0_height_ths = st.slider("height_ths", 0.1, 1., 0.5, step=0.1, \
-                    help='''height_ths (float, default = 0.5) - Maximum different in box height. \n
-Boxes wiery different text size should not be merged.''')
-            t0_width_ths = st.slider("width_ths", 0.1, 1., 0.5, step=0.1, \
-                    help="width_ths (float, default = 0.5) - Maximum horizontal \
-                          distance to merge boxes.")
-            t0_add_margin = st.slider("add_margin", 0.1, 1., 0.1, step=0.1, \
-                    help='''add_margin (float, default = 0.1) - \
-                            Extend bounding boxes in all direction by certain value. \n
-This is rtant for language with complex script (E.g. Thai).''')
-            t0_optimal_num_chars = st.slider("optimal_num_chars", None, 100, None, step=10, \
-                    help="optimal_num_chars (int, default = None) - If specified, bounding boxes \
-                          with estimated number of characters near this value are returned first.")
-
-        with col2.expander("Choose detection hyperparameters for " + reader_type_list[1], \
-                           expanded=False):
-            t1_det_algorithm = st.selectbox('det_algorithm', ['DB'], \
-            	    help='Type of detection algorithm selected. (default = DB)')
-            t1_det_max_side_len = st.slider('det_max_side_len', 500, 2000, 960, step=10, \
-                    help='''The maximum size of the long side of the image. (default = 960)\n
-Limit thximum image height and width.\n
-When theg side exceeds this value, the long side will be resized to this size, and the short side \
-will be ed proportionally.''')
-            t1_det_db_thresh =  st.slider('det_db_thresh', 0.1, 1., 0.3, step=0.1, \
-                    help='''Binarization threshold value of DB output map. (default = 0.3) \n
-Used to er the binarized image of DB prediction, setting 0.-0.3 has no obvious effect on the result.''')
-            t1_det_db_box_thresh = st.slider('det_db_box_thresh', 0.1, 1., 0.6, step=0.1, \
-                    help='''The threshold value of the DB output box. (default = 0.6) \n
-DB post-essing filter box threshold, if there is a missing box detected, it can be reduced as appropriate. \n
-Boxes sclower than this value will be discard.''')
-            t1_det_db_unclip_ratio = st.slider('det_db_unclip_ratio', 1., 3.0, 1.6, step=0.1, \
-                    help='''The expanded ratio of DB output box. (default = 1.6) \n
-Indicatee compactness of the text box, the smaller the value, the closer the text box to the text.''')
-            t1_det_east_score_thresh = st.slider('det_east_cover_thresh', 0.1, 1., 0.8, step=0.1, \
-                    help="Binarization threshold value of EAST output map. (default = 0.8)")
-            t1_det_east_cover_thresh = st.slider('det_east_cover_thresh', 0.1, 1., 0.1, step=0.1, \
-                    help='''The threshold value of the EAST output box. (default = 0.1) \n
-Boxes sclower than this value will be discarded.''')
-            t1_det_east_nms_thresh = st.slider('det_east_nms_thresh', 0.1, 1., 0.2, step=0.1, \
-                    help="The NMS threshold value of EAST model output box. (default = 0.2)")
-            t1_det_db_score_mode = st.selectbox('det_db_score_mode', ['fast', 'slow'], \
-                    help='''slow: use polygon box to calculate bbox score, fast: use rectangle box \
-                    to calculate. (default = fast) \n
-Use rectlar box to calculate faster, and polygonal box more accurate for curved text area.''')
-
-        with col2.expander("Choose detection hyperparameters for " + reader_type_list[2], \
-                           expanded=False):
-            t2_det = st.selectbox('det', ['DB_r18','DB_r50','DBPP_r50','DRRG','FCE_IC15', \
-                                          'FCE_CTW_DCNv2','MaskRCNN_CTW','MaskRCNN_IC15', \
-                                          'MaskRCNN_IC17', 'PANet_CTW','PANet_IC15','PS_CTW',\
-                                          'PS_IC15','Tesseract','TextSnake'], 10, \
-            		help='Text detection algorithm. (default = PANet_IC15)')
-            st.write("###### *More about text detection models*  👉  \
-                    [here](https://mmocr.readthedocs.io/en/latest/textdet_models.html)")
-            t2_merge_xdist = st.slider('merge_xdist', 1, 50, 20, step=1, \
-                    help='The maximum x-axis distance to merge boxes. (defaut=20)')
-
-        with col2.expander("Choose detection hyperparameters for " + reader_type_list[3], \
-                           expanded=False):
-            t3_psm = st.selectbox('Page segmentation mode (psm)', \
-                                  [' -  Default', \
-                                   ' 4  Assume a single column of text of variable sizes', \
-                                   ' 5  Assume a single uniform block of vertically aligned text', \
-                                   ' 6  Assume a single uniform block of text', \
-                                   ' 7  Treat the image as a single text line', \
-                                   ' 8  Treat the image as a single word', \
-                                   ' 9  Treat the image as a single word in a circle', \
-                                   '10  Treat the image as a single character', \
-                                   '11  Sparse text. Find as much text as possible in no \
-                                        particular order', \
-                                   '13  Raw line. Treat the image as a single text line, \
-                                        bypassing hacks that are Tesseract-specific'])
-            t3_oem = st.selectbox('OCR engine mode', ['0  Legacy engine only', \
-                                  '1  Neural nets LSTM engine only', \
-                                  '2  Legacy + LSTM engines', \
-                                  '3  Default, based on what is available'], 3)
-            t3_whitelist = st.text_input('Limit tesseract to recognize only this characters :', \
-                    placeholder='Limit tesseract to recognize only this characters', \
-                    help='Example for numbers only : 0123456789')
-
-        color_hex = col2.color_picker('Set a color for box outlines:', '#004C99')
-        color_part = color_hex.lstrip('#')
-        color = tuple(int(color_part[i:i+2], 16) for i in (0, 2, 4))
-
-        submit_detect = st.form_submit_button("Launch detection")
-
-##----------- Process text detection --------------------------------------------------------------
-    if submit_detect:
-        # Process text detection
-
-        if t0_optimal_num_chars == 0:
-            t0_optimal_num_chars = None
-
-        # Construct the config Tesseract parameter
-        t3_config = ''
-        psm = t3_psm[:2]
-        if psm != ' -':
-            t3_config += '--psm ' + psm.strip()
-        oem = t3_oem[:1]
-        if oem != '3':
-            t3_config += ' --oem ' + oem
-        if t3_whitelist != '':
-            t3_config += ' -c tessedit_char_whitelist=' + t3_whitelist
-
-        list_params_det = \
-            [[easyocr_lang, \
-              {'min_size': t0_min_size, 'text_threshold': t0_text_threshold, \
-               'low_text': t0_low_text, 'link_threshold': t0_link_threshold, \
-               'canvas_size': t0_canvas_size, 'mag_ratio': t0_mag_ratio, \
-               'slope_ths': t0_slope_ths, 'ycenter_ths': t0_ycenter_ths, \
-               'height_ths': t0_height_ths, 'width_ths': t0_width_ths, \
-               'add_margin': t0_add_margin, 'optimal_num_chars': t0_optimal_num_chars \
-              }], \
-             [ppocr_lang, \
-              {'det_algorithm': t1_det_algorithm, 'det_max_side_len': t1_det_max_side_len, \
-               'det_db_thresh': t1_det_db_thresh, 'det_db_box_thresh': t1_det_db_box_thresh, \
-               'det_db_unclip_ratio': t1_det_db_unclip_ratio, \
-               'det_east_score_thresh': t1_det_east_score_thresh, \
-               'det_east_cover_thresh': t1_det_east_cover_thresh, \
-               'det_east_nms_thresh': t1_det_east_nms_thresh, \
-               'det_db_score_mode': t1_det_db_score_mode}],
-             [mmocr_lang, {'det': t2_det, 'merge_xdist': t2_merge_xdist}],
-             [tesserocr_lang, {'lang': tesserocr_lang, 'config': t3_config}]
-            ]
-
-        show_info1 = st.empty()
-        show_info1.info("Readers initializations in progress (it may take a while) ...")
-        list_readers = init_readers(list_params_det)
-
-        show_info1.info("Text detection in progress ...")
-        list_images, list_coordinates = process_detect(image_path, list_images, list_readers, \
-                                                       list_params_det, color)
-        show_info1.empty()
-
-        # Clear previous recognition results
-        st.session_state.df_results = pd.DataFrame([])
-
-        st.session_state.list_readers = list_readers
-        st.session_state.list_coordinates = list_coordinates
-        st.session_state.list_images = list_images
-        st.session_state.list_params_det = list_params_det
-
-        if 'columns_size' not in st.session_state:
-            st.session_state.columns_size = [2] + [1 for x in reader_type_list[1:]]
-        if 'column_width' not in st.session_state:
-            st.session_state.column_width = [500] + [400 for x in reader_type_list[1:]]
-        if 'columns_color' not in st.session_state:
-            st.session_state.columns_color = ["rgb(228,26,28)"] + \
-                                             ["rgb(0,0,0)" for x in reader_type_list[1:]]
-
-    if st.session_state.list_coordinates:
-        list_coordinates = st.session_state.list_coordinates
-        list_images = st.session_state.list_images
-        list_readers = st.session_state.list_readers
-        list_params_det = st.session_state.list_params_det
-
-##----------- Text detection results --------------------------------------------------------------
-        st.subheader("Text detection")
-        show_detect = st.empty()
-        list_ok_detect = []
-        with show_detect.container():
-            columns = st.columns(st.session_state.columns_size, ) #gap='medium')
-            for no_col, col in enumerate(columns):
-                column_title = '<p style="font-size: 20px;color:' + \
-                               st.session_state.columns_color[no_col] + \
-                               ';">Detection with ' + reader_type_list[no_col]+ '</p>'
-                col.markdown(column_title, unsafe_allow_html=True)
-                if isinstance(list_images[no_col+2], PIL.Image.Image):
-                    col.image(list_images[no_col+2], width=st.session_state.column_width[no_col], \
-                              use_column_width=True)
-                    list_ok_detect.append(reader_type_list[no_col])
-                else:
-                    col.write(list_images[no_col+2], use_column_width=True)
-
-        st.subheader("Text recognition")
+    ##----------- Process input image -----------------------------------------------------------------
+    if image_file is not None:
+        image_path, image_orig, image_cv2 = load_image(image_file)
+        list_images = [image_orig, image_cv2]
 
-        st.markdown("##### Using detection performed above by:")
-        st.radio('Choose the detecter:', list_ok_detect, key='detect_reader', \
-                 horizontal=True, on_change=highlight)
+    ##----------- Form with original image & hyperparameters for detectors ----------------------------
+        with st.form("form1"):
+            col1, col2 = st.columns(2, ) #gap="medium")
+            col1.markdown("##### Original image")
+            col1.image(list_images[0], width=500, use_column_width=True)
+            col2.markdown("##### Hyperparameters values for detection")
 
-##----------- Form with hyperparameters for recognition -----------------------
-        st.markdown("##### Hyperparameters values for recognition:")
-        with st.form("form2"):
-            with st.expander("Choose recognition hyperparameters for " + reader_type_list[0], \
+            with col2.expander("Choose detection hyperparameters for " + reader_type_list[0], \
                             expanded=False):
-                t0_decoder = st.selectbox('decoder', ['greedy', 'beamsearch', 'wordbeamsearch'], \
-                    help="decoder (string, default = 'greedy') - options are 'greedy', \
-                         'beamsearch' and 'wordbeamsearch.")
-                t0_beamWidth = st.slider('beamWidth', 2, 20, 5, step=1, \
-                    help="beamWidth (int, default = 5) - How many beam to keep when decoder = \
-                         'beamsearch' or 'wordbeamsearch'.")
-                t0_batch_size = st.slider('batch_size', 1, 10, 1, step=1, \
-                    help="batch_size (int, default = 1) - batch_size>1 will make EasyOCR faster \
-                          but use more memory.")
-                t0_workers = st.slider('workers', 0, 10, 0, step=1, \
-                    help="workers (int, default = 0) - Number thread used in of dataloader.")
-                t0_allowlist = st.text_input('allowlist', value="", max_chars=None, \
-                    placeholder='Force EasyOCR to recognize only this subset of characters', \
-                    help='''allowlist (string) - Force EasyOCR to recognize only subset of characters.\n
-        Usefor specific problem (E.g. license plate, etc.)''')
-                t0_blocklist = st.text_input('blocklist', value="", max_chars=None, \
-                    placeholder='Block subset of character (will be ignored if allowlist is given)', \
-                    help='''blocklist (string) - Block subset of character. This argument will be \
-                         ignored if allowlist is given.''')
-                t0_detail = st.radio('detail', [0, 1], 1, horizontal=True, \
-                    help="detail (int, default = 1) - Set this to 0 for simple output")
-                t0_paragraph = st.radio('paragraph', [True, False], 1, horizontal=True, \
-                    help='paragraph (bool, default = False) - Combine result into paragraph')
-                t0_contrast_ths = st.slider('contrast_ths', 0.05, 1., 0.1, step=0.01, \
-                    help='''contrast_ths (float, default = 0.1) - Text box with contrast lower than \
-                         this value will be passed into model 2 times.\n
-        Firs with original image and second with contrast adjusted to 'adjust_contrast' value.\n
-        The with more confident level will be returned as a result.''')
-                t0_adjust_contrast = st.slider('adjust_contrast', 0.1, 1., 0.5, step=0.1, \
-                    help = 'adjust_contrast (float, default = 0.5) - target contrast level for low \
-                    contrast text box')
-
-            with st.expander("Choose recognition hyperparameters for " + reader_type_list[1], \
+                t0_min_size = st.slider("min_size", 1, 20, 10, step=1, \
+                        help="min_size (int, default = 10) - Filter text box smaller than \
+                            minimum value in pixel")
+                t0_text_threshold = st.slider("text_threshold", 0.1, 1., 0.7, step=0.1, \
+                        help="text_threshold (float, default = 0.7) - Text confidence threshold")
+                t0_low_text = st.slider("low_text", 0.1, 1., 0.4, step=0.1, \
+                        help="low_text (float, default = 0.4) - Text low-bound score")
+                t0_link_threshold = st.slider("link_threshold", 0.1, 1., 0.4, step=0.1, \
+                        help="link_threshold (float, default = 0.4) - Link confidence threshold")
+                t0_canvas_size = st.slider("canvas_size", 2000, 5000, 2560, step=10, \
+                        help='''canvas_size (int, default = 2560) \n
+    Maximum e size. Image bigger than this value will be resized down''')
+                t0_mag_ratio = st.slider("mag_ratio", 0.1, 5., 1., step=0.1, \
+                        help="mag_ratio (float, default = 1) - Image magnification ratio")
+                t0_slope_ths = st.slider("slope_ths", 0.01, 1., 0.1, step=0.01, \
+                        help='''slope_ths (float, default = 0.1) - Maximum slope \
+                                (delta y/delta x) to considered merging. \n
+    Low valuans tiled boxes will not be merged.''')
+                t0_ycenter_ths = st.slider("ycenter_ths", 0.1, 1., 0.5, step=0.1, \
+                        help='''ycenter_ths (float, default = 0.5) - Maximum shift in y direction. \n
+    Boxes wiifferent level should not be merged.''')
+                t0_height_ths = st.slider("height_ths", 0.1, 1., 0.5, step=0.1, \
+                        help='''height_ths (float, default = 0.5) - Maximum different in box height. \n
+    Boxes wiery different text size should not be merged.''')
+                t0_width_ths = st.slider("width_ths", 0.1, 1., 0.5, step=0.1, \
+                        help="width_ths (float, default = 0.5) - Maximum horizontal \
+                            distance to merge boxes.")
+                t0_add_margin = st.slider("add_margin", 0.1, 1., 0.1, step=0.1, \
+                        help='''add_margin (float, default = 0.1) - \
+                                Extend bounding boxes in all direction by certain value. \n
+    This is rtant for language with complex script (E.g. Thai).''')
+                t0_optimal_num_chars = st.slider("optimal_num_chars", None, 100, None, step=10, \
+                        help="optimal_num_chars (int, default = None) - If specified, bounding boxes \
+                            with estimated number of characters near this value are returned first.")
+
+            with col2.expander("Choose detection hyperparameters for " + reader_type_list[1], \
                             expanded=False):
-                t1_rec_algorithm = st.selectbox('rec_algorithm', ['CRNN', 'SVTR_LCNet'], 0, \
-                    help="Type of recognition algorithm selected. (default=CRNN)")
-                t1_rec_batch_num = st.slider('rec_batch_num', 1, 50, step=1, \
-                    help="When performing recognition, the batchsize of forward images. \
-                         (default=30)")
-                t1_max_text_length = st.slider('max_text_length', 3, 250, 25, step=1, \
-                    help="The maximum text length that the recognition algorithm can recognize. \
-                         (default=25)")
-                t1_use_space_char = st.radio('use_space_char', [True, False], 0, horizontal=True, \
-                    help="Whether to recognize spaces. (default=TRUE)")
-                t1_drop_score = st.slider('drop_score', 0., 1., 0.25, step=.05, \
-                    help="Filter the output by score (from the recognition model), and those \
-                          below this score will not be returned. (default=0.5)")
-
-            with st.expander("Choose recognition hyperparameters for " + reader_type_list[2], \
+                t1_det_algorithm = st.selectbox('det_algorithm', ['DB'], \
+                        help='Type of detection algorithm selected. (default = DB)')
+                t1_det_max_side_len = st.slider('det_max_side_len', 500, 2000, 960, step=10, \
+                        help='''The maximum size of the long side of the image. (default = 960)\n
+    Limit thximum image height and width.\n
+    When theg side exceeds this value, the long side will be resized to this size, and the short side \
+    will be ed proportionally.''')
+                t1_det_db_thresh =  st.slider('det_db_thresh', 0.1, 1., 0.3, step=0.1, \
+                        help='''Binarization threshold value of DB output map. (default = 0.3) \n
+    Used to er the binarized image of DB prediction, setting 0.-0.3 has no obvious effect on the result.''')
+                t1_det_db_box_thresh = st.slider('det_db_box_thresh', 0.1, 1., 0.6, step=0.1, \
+                        help='''The threshold value of the DB output box. (default = 0.6) \n
+    DB post-essing filter box threshold, if there is a missing box detected, it can be reduced as appropriate. \n
+    Boxes sclower than this value will be discard.''')
+                t1_det_db_unclip_ratio = st.slider('det_db_unclip_ratio', 1., 3.0, 1.6, step=0.1, \
+                        help='''The expanded ratio of DB output box. (default = 1.6) \n
+    Indicatee compactness of the text box, the smaller the value, the closer the text box to the text.''')
+                t1_det_east_score_thresh = st.slider('det_east_cover_thresh', 0.1, 1., 0.8, step=0.1, \
+                        help="Binarization threshold value of EAST output map. (default = 0.8)")
+                t1_det_east_cover_thresh = st.slider('det_east_cover_thresh', 0.1, 1., 0.1, step=0.1, \
+                        help='''The threshold value of the EAST output box. (default = 0.1) \n
+    Boxes sclower than this value will be discarded.''')
+                t1_det_east_nms_thresh = st.slider('det_east_nms_thresh', 0.1, 1., 0.2, step=0.1, \
+                        help="The NMS threshold value of EAST model output box. (default = 0.2)")
+                t1_det_db_score_mode = st.selectbox('det_db_score_mode', ['fast', 'slow'], \
+                        help='''slow: use polygon box to calculate bbox score, fast: use rectangle box \
+                        to calculate. (default = fast) \n
+    Use rectlar box to calculate faster, and polygonal box more accurate for curved text area.''')
+
+            with col2.expander("Choose detection hyperparameters for " + reader_type_list[2], \
                             expanded=False):
-                t2_recog = st.selectbox('recog', ['ABINet','CRNN','CRNN_TPS','MASTER', \
-                              'NRTR_1/16-1/8','NRTR_1/8-1/4','RobustScanner','SAR','SAR_CN', \
-                              'SATRN','SATRN_sm','SEG','Tesseract'], 7, \
-                        help='Text recognition algorithm. (default = SAR)')
-                st.write("###### *More about text recognition models*  👉  \
-                        [here](https://mmocr.readthedocs.io/en/latest/textrecog_models.html)")
-
-            with st.expander("Choose recognition hyperparameters for " + reader_type_list[3], \
+                t2_det = st.selectbox('det', ['DB_r18','DB_r50','DBPP_r50','DRRG','FCE_IC15', \
+                                            'FCE_CTW_DCNv2','MaskRCNN_CTW','MaskRCNN_IC15', \
+                                            'MaskRCNN_IC17', 'PANet_CTW','PANet_IC15','PS_CTW',\
+                                            'PS_IC15','Tesseract','TextSnake'], 10, \
+                        help='Text detection algorithm. (default = PANet_IC15)')
+                st.write("###### *More about text detection models*  👉  \
+                        [here](https://mmocr.readthedocs.io/en/latest/textdet_models.html)")
+                t2_merge_xdist = st.slider('merge_xdist', 1, 50, 20, step=1, \
+                        help='The maximum x-axis distance to merge boxes. (defaut=20)')
+
+            with col2.expander("Choose detection hyperparameters for " + reader_type_list[3], \
                             expanded=False):
-                t3r_psm = st.selectbox('Page segmentation mode (psm)', \
-                                       [' -  Default', \
-                                       ' 4  Assume a single column of text of variable sizes', \
-                                       ' 5  Assume a single uniform block of vertically aligned \
-                                            text', \
-                                       ' 6  Assume a single uniform block of text', \
-                                       ' 7  Treat the image as a single text line', \
-                                       ' 8  Treat the image as a single word', \
-                                       ' 9  Treat the image as a single word in a circle', \
-                                       '10  Treat the image as a single character', \
-                                       '11  Sparse text. Find as much text as possible in no \
+                t3_psm = st.selectbox('Page segmentation mode (psm)', \
+                                    [' -  Default', \
+                                    ' 4  Assume a single column of text of variable sizes', \
+                                    ' 5  Assume a single uniform block of vertically aligned text', \
+                                    ' 6  Assume a single uniform block of text', \
+                                    ' 7  Treat the image as a single text line', \
+                                    ' 8  Treat the image as a single word', \
+                                    ' 9  Treat the image as a single word in a circle', \
+                                    '10  Treat the image as a single character', \
+                                    '11  Sparse text. Find as much text as possible in no \
                                             particular order', \
-                                       '13  Raw line. Treat the image as a single text line, \
+                                    '13  Raw line. Treat the image as a single text line, \
                                             bypassing hacks that are Tesseract-specific'])
-                t3r_oem = st.selectbox('OCR engine mode', ['0  Legacy engine only', \
-                                       '1  Neural nets LSTM engine only', \
-                                       '2  Legacy + LSTM engines', \
-                                       '3  Default, based on what is available'], 3)
-                t3r_whitelist = st.text_input('Limit tesseract to recognize only this \
-                                    characters :', \
-                                    placeholder='Limit tesseract to recognize only this characters', \
-                                    help='Example for numbers only : 0123456789')
-
-            submit_reco = st.form_submit_button("Launch recognition")
-
-        if submit_reco:
-            process_detect.clear()
-##----------- Process recognition ------------------------------------------
-            reader_ind = reader_type_dict[st.session_state.detect_reader]
-            list_boxes = list_coordinates[reader_ind]
+                t3_oem = st.selectbox('OCR engine mode', ['0  Legacy engine only', \
+                                    '1  Neural nets LSTM engine only', \
+                                    '2  Legacy + LSTM engines', \
+                                    '3  Default, based on what is available'], 3)
+                t3_whitelist = st.text_input('Limit tesseract to recognize only this characters :', \
+                        placeholder='Limit tesseract to recognize only this characters', \
+                        help='Example for numbers only : 0123456789')
+
+            color_hex = col2.color_picker('Set a color for box outlines:', '#004C99')
+            color_part = color_hex.lstrip('#')
+            color = tuple(int(color_part[i:i+2], 16) for i in (0, 2, 4))
+
+            submit_detect = st.form_submit_button("Launch detection")
+
+    ##----------- Process text detection --------------------------------------------------------------
+        if submit_detect:
+            # Process text detection
+
+            if t0_optimal_num_chars == 0:
+                t0_optimal_num_chars = None
 
             # Construct the config Tesseract parameter
-            t3r_config = ''
-            psm = t3r_psm[:2]
+            t3_config = ''
+            psm = t3_psm[:2]
             if psm != ' -':
-                t3r_config += '--psm ' + psm.strip()
-            oem = t3r_oem[:1]
+                t3_config += '--psm ' + psm.strip()
+            oem = t3_oem[:1]
             if oem != '3':
-                t3r_config += ' --oem ' + oem
-            if t3r_whitelist != '':
-                t3r_config += ' -c tessedit_char_whitelist=' + t3r_whitelist
-
-            list_params_rec = \
-                [{'decoder': t0_decoder, 'beamWidth': t0_beamWidth, \
-                  'batch_size': t0_batch_size, 'workers': t0_workers, \
-                  'allowlist': t0_allowlist, 'blocklist': t0_blocklist, \
-                  'detail': t0_detail, 'paragraph': t0_paragraph, \
-                  'contrast_ths': t0_contrast_ths, 'adjust_contrast': t0_adjust_contrast
-                 },
-                 { **list_params_det[1][1], **{'rec_algorithm': t1_rec_algorithm, \
-                 'rec_batch_num': t1_rec_batch_num, 'max_text_length': t1_max_text_length, \
-                  'use_space_char': t1_use_space_char, 'drop_score': t1_drop_score}, \
-                  **{'lang': list_params_det[1][0]}
-                 },
-                 {'recog': t2_recog},
-                 {'lang': tesserocr_lang, 'config': t3r_config}
+                t3_config += ' --oem ' + oem
+            if t3_whitelist != '':
+                t3_config += ' -c tessedit_char_whitelist=' + t3_whitelist
+
+            list_params_det = \
+                [[easyocr_lang, \
+                {'min_size': t0_min_size, 'text_threshold': t0_text_threshold, \
+                'low_text': t0_low_text, 'link_threshold': t0_link_threshold, \
+                'canvas_size': t0_canvas_size, 'mag_ratio': t0_mag_ratio, \
+                'slope_ths': t0_slope_ths, 'ycenter_ths': t0_ycenter_ths, \
+                'height_ths': t0_height_ths, 'width_ths': t0_width_ths, \
+                'add_margin': t0_add_margin, 'optimal_num_chars': t0_optimal_num_chars \
+                }], \
+                [ppocr_lang, \
+                {'det_algorithm': t1_det_algorithm, 'det_max_side_len': t1_det_max_side_len, \
+                'det_db_thresh': t1_det_db_thresh, 'det_db_box_thresh': t1_det_db_box_thresh, \
+                'det_db_unclip_ratio': t1_det_db_unclip_ratio, \
+                'det_east_score_thresh': t1_det_east_score_thresh, \
+                'det_east_cover_thresh': t1_det_east_cover_thresh, \
+                'det_east_nms_thresh': t1_det_east_nms_thresh, \
+                'det_db_score_mode': t1_det_db_score_mode}],
+                [mmocr_lang, {'det': t2_det, 'merge_xdist': t2_merge_xdist}],
+                [tesserocr_lang, {'lang': tesserocr_lang, 'config': t3_config}]
                 ]
 
-            show_info2 = st.empty()
-
-            with show_info2.container():
-                st.info("Text recognition in progress ...")
-                df_results, df_results_tesseract, list_reco_status = \
-                        process_recog(list_readers, list_images[1], list_boxes, list_params_rec)
-            show_info2.empty()
-
-            st.session_state.df_results = df_results
-            st.session_state.list_boxes = list_boxes
-            st.session_state.df_results_tesseract = df_results_tesseract
-            st.session_state.list_reco_status = list_reco_status
-
-        if 'df_results' in st.session_state:
-            if not st.session_state.df_results.empty:
-##----------- Show recognition results ------------------------------------------------------------
-                results_cols = st.session_state.df_results.columns
-                list_col_text = np.arange(1, len(cols_size), 2)
-                list_col_confid = np.arange(2, len(cols_size), 2)
-
-                dict_draw_reco = {'in_image': st.session_state.list_images[1], \
-                                'in_boxes_coordinates': st.session_state.list_boxes, \
-                                'in_list_texts': [st.session_state.df_results[x].to_list() \
-                                                    for x in results_cols[list_col_text]], \
-                                'in_list_confid': [st.session_state.df_results[x].to_list() \
-                                                    for x in results_cols[list_col_confid]], \
-                                'in_dict_back_colors': dict_back_colors, \
-                                'in_df_results_tesseract' : st.session_state.df_results_tesseract, \
-                                'in_reader_type_list': reader_type_list
-                                }
-                show_reco = st.empty()
-
-                with st.form("form3"):
-                    st.plotly_chart(fig_colorscale, use_container_width=True)
-
-                    col_font, col_threshold = st.columns(2)
-
-                    col_font.slider('Font scale', 1, 7, 1, step=1, key="font_scale_sld")
-                    col_threshold.slider('% confidence threshold for text color change', 40, 100, 64, \
-                                        step=1, key="conf_threshold_sld")
-                    col_threshold.write("(text color is black below this % confidence threshold, \
-                                        and white above)")
-
-                    draw_reco_images(**dict_draw_reco)
-
-                    submit_resize = st.form_submit_button("Refresh")
-
-                if submit_resize:
-                    draw_reco_images(**dict_draw_reco, \
-                                    in_font_scale=st.session_state.font_scale_sld, \
-                                    in_conf_threshold=st.session_state.conf_threshold_sld)
-
-                st.subheader("Recognition details")
-                with st.expander("Detailed areas for EasyOCR, PPOCR, MMOCR", expanded=True):
-                    cols = st.columns(cols_size)
-                    cols[0].markdown('#### Detected area')
-                    for i in range(1, (len(reader_type_list)-1)*2, 2):
-                        cols[i].markdown('#### with ' + reader_type_list[i//2])
-
-                    for row in st.session_state.df_results.itertuples():
-                        #cols = st.columns(1 + len(reader_type_list)*2)
+            show_info1 = st.empty()
+            show_info1.info("Readers initializations in progress (it may take a while) ...")
+            list_readers = init_readers(list_params_det)
+
+            show_info1.info("Text detection in progress ...")
+            list_images, list_coordinates = process_detect(image_path, list_images, list_readers, \
+                                                        list_params_det, color)
+            show_info1.empty()
+
+            # Clear previous recognition results
+            st.session_state.df_results = pd.DataFrame([])
+
+            st.session_state.list_readers = list_readers
+            st.session_state.list_coordinates = list_coordinates
+            st.session_state.list_images = list_images
+            st.session_state.list_params_det = list_params_det
+
+            if 'columns_size' not in st.session_state:
+                st.session_state.columns_size = [2] + [1 for x in reader_type_list[1:]]
+            if 'column_width' not in st.session_state:
+                st.session_state.column_width = [500] + [400 for x in reader_type_list[1:]]
+            if 'columns_color' not in st.session_state:
+                st.session_state.columns_color = ["rgb(228,26,28)"] + \
+                                                ["rgb(0,0,0)" for x in reader_type_list[1:]]
+
+        if st.session_state.list_coordinates:
+            list_coordinates = st.session_state.list_coordinates
+            list_images = st.session_state.list_images
+            list_readers = st.session_state.list_readers
+            list_params_det = st.session_state.list_params_det
+
+    ##----------- Text detection results --------------------------------------------------------------
+            st.subheader("Text detection")
+            show_detect = st.empty()
+            list_ok_detect = []
+            with show_detect.container():
+                columns = st.columns(st.session_state.columns_size, ) #gap='medium')
+                for no_col, col in enumerate(columns):
+                    column_title = '<p style="font-size: 20px;color:' + \
+                                st.session_state.columns_color[no_col] + \
+                                ';">Detection with ' + reader_type_list[no_col]+ '</p>'
+                    col.markdown(column_title, unsafe_allow_html=True)
+                    if isinstance(list_images[no_col+2], PIL.Image.Image):
+                        col.image(list_images[no_col+2], width=st.session_state.column_width[no_col], \
+                                use_column_width=True)
+                        list_ok_detect.append(reader_type_list[no_col])
+                    else:
+                        col.write(list_images[no_col+2], use_column_width=True)
+
+            st.subheader("Text recognition")
+
+            st.markdown("##### Using detection performed above by:")
+            st.radio('Choose the detecter:', list_ok_detect, key='detect_reader', \
+                    horizontal=True, on_change=highlight)
+
+    ##----------- Form with hyperparameters for recognition -----------------------
+            st.markdown("##### Hyperparameters values for recognition:")
+            with st.form("form2"):
+                with st.expander("Choose recognition hyperparameters for " + reader_type_list[0], \
+                                expanded=False):
+                    t0_decoder = st.selectbox('decoder', ['greedy', 'beamsearch', 'wordbeamsearch'], \
+                        help="decoder (string, default = 'greedy') - options are 'greedy', \
+                            'beamsearch' and 'wordbeamsearch.")
+                    t0_beamWidth = st.slider('beamWidth', 2, 20, 5, step=1, \
+                        help="beamWidth (int, default = 5) - How many beam to keep when decoder = \
+                            'beamsearch' or 'wordbeamsearch'.")
+                    t0_batch_size = st.slider('batch_size', 1, 10, 1, step=1, \
+                        help="batch_size (int, default = 1) - batch_size>1 will make EasyOCR faster \
+                            but use more memory.")
+                    t0_workers = st.slider('workers', 0, 10, 0, step=1, \
+                        help="workers (int, default = 0) - Number thread used in of dataloader.")
+                    t0_allowlist = st.text_input('allowlist', value="", max_chars=None, \
+                        placeholder='Force EasyOCR to recognize only this subset of characters', \
+                        help='''allowlist (string) - Force EasyOCR to recognize only subset of characters.\n
+            Usefor specific problem (E.g. license plate, etc.)''')
+                    t0_blocklist = st.text_input('blocklist', value="", max_chars=None, \
+                        placeholder='Block subset of character (will be ignored if allowlist is given)', \
+                        help='''blocklist (string) - Block subset of character. This argument will be \
+                            ignored if allowlist is given.''')
+                    t0_detail = st.radio('detail', [0, 1], 1, horizontal=True, \
+                        help="detail (int, default = 1) - Set this to 0 for simple output")
+                    t0_paragraph = st.radio('paragraph', [True, False], 1, horizontal=True, \
+                        help='paragraph (bool, default = False) - Combine result into paragraph')
+                    t0_contrast_ths = st.slider('contrast_ths', 0.05, 1., 0.1, step=0.01, \
+                        help='''contrast_ths (float, default = 0.1) - Text box with contrast lower than \
+                            this value will be passed into model 2 times.\n
+            Firs with original image and second with contrast adjusted to 'adjust_contrast' value.\n
+            The with more confident level will be returned as a result.''')
+                    t0_adjust_contrast = st.slider('adjust_contrast', 0.1, 1., 0.5, step=0.1, \
+                        help = 'adjust_contrast (float, default = 0.5) - target contrast level for low \
+                        contrast text box')
+
+                with st.expander("Choose recognition hyperparameters for " + reader_type_list[1], \
+                                expanded=False):
+                    t1_rec_algorithm = st.selectbox('rec_algorithm', ['CRNN', 'SVTR_LCNet'], 0, \
+                        help="Type of recognition algorithm selected. (default=CRNN)")
+                    t1_rec_batch_num = st.slider('rec_batch_num', 1, 50, step=1, \
+                        help="When performing recognition, the batchsize of forward images. \
+                            (default=30)")
+                    t1_max_text_length = st.slider('max_text_length', 3, 250, 25, step=1, \
+                        help="The maximum text length that the recognition algorithm can recognize. \
+                            (default=25)")
+                    t1_use_space_char = st.radio('use_space_char', [True, False], 0, horizontal=True, \
+                        help="Whether to recognize spaces. (default=TRUE)")
+                    t1_drop_score = st.slider('drop_score', 0., 1., 0.25, step=.05, \
+                        help="Filter the output by score (from the recognition model), and those \
+                            below this score will not be returned. (default=0.5)")
+
+                with st.expander("Choose recognition hyperparameters for " + reader_type_list[2], \
+                                expanded=False):
+                    t2_recog = st.selectbox('recog', ['ABINet','CRNN','CRNN_TPS','MASTER', \
+                                'NRTR_1/16-1/8','NRTR_1/8-1/4','RobustScanner','SAR','SAR_CN', \
+                                'SATRN','SATRN_sm','SEG','Tesseract'], 7, \
+                            help='Text recognition algorithm. (default = SAR)')
+                    st.write("###### *More about text recognition models*  👉  \
+                            [here](https://mmocr.readthedocs.io/en/latest/textrecog_models.html)")
+
+                with st.expander("Choose recognition hyperparameters for " + reader_type_list[3], \
+                                expanded=False):
+                    t3r_psm = st.selectbox('Page segmentation mode (psm)', \
+                                        [' -  Default', \
+                                        ' 4  Assume a single column of text of variable sizes', \
+                                        ' 5  Assume a single uniform block of vertically aligned \
+                                                text', \
+                                        ' 6  Assume a single uniform block of text', \
+                                        ' 7  Treat the image as a single text line', \
+                                        ' 8  Treat the image as a single word', \
+                                        ' 9  Treat the image as a single word in a circle', \
+                                        '10  Treat the image as a single character', \
+                                        '11  Sparse text. Find as much text as possible in no \
+                                                particular order', \
+                                        '13  Raw line. Treat the image as a single text line, \
+                                                bypassing hacks that are Tesseract-specific'])
+                    t3r_oem = st.selectbox('OCR engine mode', ['0  Legacy engine only', \
+                                        '1  Neural nets LSTM engine only', \
+                                        '2  Legacy + LSTM engines', \
+                                        '3  Default, based on what is available'], 3)
+                    t3r_whitelist = st.text_input('Limit tesseract to recognize only this \
+                                        characters :', \
+                                        placeholder='Limit tesseract to recognize only this characters', \
+                                        help='Example for numbers only : 0123456789')
+
+                submit_reco = st.form_submit_button("Launch recognition")
+
+            if submit_reco:
+                process_detect.clear()
+    ##----------- Process recognition ------------------------------------------
+                reader_ind = reader_type_dict[st.session_state.detect_reader]
+                list_boxes = list_coordinates[reader_ind]
+
+                # Construct the config Tesseract parameter
+                t3r_config = ''
+                psm = t3r_psm[:2]
+                if psm != ' -':
+                    t3r_config += '--psm ' + psm.strip()
+                oem = t3r_oem[:1]
+                if oem != '3':
+                    t3r_config += ' --oem ' + oem
+                if t3r_whitelist != '':
+                    t3r_config += ' -c tessedit_char_whitelist=' + t3r_whitelist
+
+                list_params_rec = \
+                    [{'decoder': t0_decoder, 'beamWidth': t0_beamWidth, \
+                    'batch_size': t0_batch_size, 'workers': t0_workers, \
+                    'allowlist': t0_allowlist, 'blocklist': t0_blocklist, \
+                    'detail': t0_detail, 'paragraph': t0_paragraph, \
+                    'contrast_ths': t0_contrast_ths, 'adjust_contrast': t0_adjust_contrast
+                    },
+                    { **list_params_det[1][1], **{'rec_algorithm': t1_rec_algorithm, \
+                    'rec_batch_num': t1_rec_batch_num, 'max_text_length': t1_max_text_length, \
+                    'use_space_char': t1_use_space_char, 'drop_score': t1_drop_score}, \
+                    **{'lang': list_params_det[1][0]}
+                    },
+                    {'recog': t2_recog},
+                    {'lang': tesserocr_lang, 'config': t3r_config}
+                    ]
+
+                show_info2 = st.empty()
+
+                with show_info2.container():
+                    st.info("Text recognition in progress ...")
+                    df_results, df_results_tesseract, list_reco_status = \
+                            process_recog(list_readers, list_images[1], list_boxes, list_params_rec)
+                show_info2.empty()
+
+                st.session_state.df_results = df_results
+                st.session_state.list_boxes = list_boxes
+                st.session_state.df_results_tesseract = df_results_tesseract
+                st.session_state.list_reco_status = list_reco_status
+
+            if 'df_results' in st.session_state:
+                if not st.session_state.df_results.empty:
+    ##----------- Show recognition results ------------------------------------------------------------
+                    results_cols = st.session_state.df_results.columns
+                    list_col_text = np.arange(1, len(cols_size), 2)
+                    list_col_confid = np.arange(2, len(cols_size), 2)
+
+                    dict_draw_reco = {'in_image': st.session_state.list_images[1], \
+                                    'in_boxes_coordinates': st.session_state.list_boxes, \
+                                    'in_list_texts': [st.session_state.df_results[x].to_list() \
+                                                        for x in results_cols[list_col_text]], \
+                                    'in_list_confid': [st.session_state.df_results[x].to_list() \
+                                                        for x in results_cols[list_col_confid]], \
+                                    'in_dict_back_colors': dict_back_colors, \
+                                    'in_df_results_tesseract' : st.session_state.df_results_tesseract, \
+                                    'in_reader_type_list': reader_type_list
+                                    }
+                    show_reco = st.empty()
+
+                    with st.form("form3"):
+                        st.plotly_chart(fig_colorscale, use_container_width=True)
+
+                        col_font, col_threshold = st.columns(2)
+
+                        col_font.slider('Font scale', 1, 7, 1, step=1, key="font_scale_sld")
+                        col_threshold.slider('% confidence threshold for text color change', 40, 100, 64, \
+                                            step=1, key="conf_threshold_sld")
+                        col_threshold.write("(text color is black below this % confidence threshold, \
+                                            and white above)")
+
+                        draw_reco_images(**dict_draw_reco)
+
+                        submit_resize = st.form_submit_button("Refresh")
+
+                    if submit_resize:
+                        draw_reco_images(**dict_draw_reco, \
+                                        in_font_scale=st.session_state.font_scale_sld, \
+                                        in_conf_threshold=st.session_state.conf_threshold_sld)
+
+                    st.subheader("Recognition details")
+                    with st.expander("Detailed areas for EasyOCR, PPOCR, MMOCR", expanded=True):
                         cols = st.columns(cols_size)
-                        cols[0].image(row.cropped_image, width=150)
-                        for ind_col in range(1, len(cols), 2):
-                            cols[ind_col].write(getattr(row, results_cols[ind_col]))
-                            cols[ind_col+1].write("("+str( \
-                                getattr(row, results_cols[ind_col+1]))+"%)")
-
-                    st.download_button(
-                        label="Download results as CSV file",
-                        data=convert_df(st.session_state.df_results),
-                        file_name='OCR_comparator_results.csv',
-                        mime='text/csv',
-                    )
-
-                if not st.session_state.df_results_tesseract.empty:
-                    with st.expander("Detailed areas for Tesseract", expanded=False):
-                        cols = st.columns([2,2,1])
                         cols[0].markdown('#### Detected area')
-                        cols[1].markdown('#### with Tesseract')
-
-                        for row in st.session_state.df_results_tesseract.itertuples():
-                            cols = st.columns([2,2,1])
-                            cols[0].image(row.cropped, width=150)
-                            cols[1].write(getattr(row, 'text'))
-                            cols[2].write("("+str(getattr(row, 'conf'))+"%)")
+                        for i in range(1, (len(reader_type_list)-1)*2, 2):
+                            cols[i].markdown('#### with ' + reader_type_list[i//2])
+
+                        for row in st.session_state.df_results.itertuples():
+                            #cols = st.columns(1 + len(reader_type_list)*2)
+                            cols = st.columns(cols_size)
+                            cols[0].image(row.cropped_image, width=150)
+                            for ind_col in range(1, len(cols), 2):
+                                cols[ind_col].write(getattr(row, results_cols[ind_col]))
+                                cols[ind_col+1].write("("+str( \
+                                    getattr(row, results_cols[ind_col+1]))+"%)")
 
                         st.download_button(
-                            label="Download Tesseract results as CSV file",
+                            label="Download results as CSV file",
                             data=convert_df(st.session_state.df_results),
-                            file_name='OCR_comparator_Tesseract_results.csv',
+                            file_name='OCR_comparator_results.csv',
                             mime='text/csv',
                         )
+
+                    if not st.session_state.df_results_tesseract.empty:
+                        with st.expander("Detailed areas for Tesseract", expanded=False):
+                            cols = st.columns([2,2,1])
+                            cols[0].markdown('#### Detected area')
+                            cols[1].markdown('#### with Tesseract')
+
+                            for row in st.session_state.df_results_tesseract.itertuples():
+                                cols = st.columns([2,2,1])
+                                cols[0].image(row.cropped, width=150)
+                                cols[1].write(getattr(row, 'text'))
+                                cols[2].write("("+str(getattr(row, 'conf'))+"%)")
+
+                            st.download_button(
+                                label="Download Tesseract results as CSV file",
+                                data=convert_df(st.session_state.df_results),
+                                file_name='OCR_comparator_Tesseract_results.csv',
+                                mime='text/csv',
+                            )

multipage.py

@@ -0,0 +1,68 @@
+"""
+This file is the framework for generating multiple Streamlit applications
+through an object oriented framework.
+
+Source: https://huggingface.co/spaces/deepset/wikipedia-assistant/tree/main
+"""
+
+# Import necessary libraries
+import streamlit as st
+from streamlit_option_menu import option_menu
+
+
+# Define the multipage class to manage the multiple apps in our program
+class MultiPage:
+    """Framework for combining multiple streamlit applications."""
+
+    def __init__(self) -> None:
+        """Constructor class to generate a list which will store all our applications as an instance variable."""
+        self.pages = []
+
+    def add_page(self, title, icon, func) -> None:
+        """Class Method to Add pages to the project
+
+        Args:
+            title ([str]): The title of page which we are adding to the list of apps
+
+            func: Python function to render this page in Streamlit
+        """
+
+        self.pages.append(
+            {
+                "title": title,
+                "icon": icon,
+                "function": func
+            }
+        )
+
+    def run(self):
+        # Drodown to select the page to run
+        st.markdown("""
+            <style>
+                section[data-testid="stSidebar"] > div:first-of-type {
+                    background-color: var(--secondary-background-color);
+                    background: var(--secondary-background-color);
+                    width: 250px;
+                    padding: 4rem 0;
+                    box-shadow: -2rem 0px 2rem 2rem rgba(0,0,0,0.16);
+                }
+                section[aria-expanded="true"] > div:nth-of-type(2) {
+                    display: none;
+                }
+                .main > div:first-of-type {
+                    padding: 1rem 0;
+                }
+            </style>
+        """, unsafe_allow_html=True)
+
+        with st.sidebar:
+            selected = option_menu("OCR Comparator",
+                                   [page["title"] for page in self.pages],
+                                   icons=[page["icon"] for page in self.pages],
+                                   menu_icon="app-indicator", default_index=0)
+
+        # Run the selected page
+        for index, item in enumerate(self.pages):
+            if item["title"] == selected:
+                self.pages[index]["function"]()
+                break

pages/About.py

@@ -1,37 +0,0 @@
-import streamlit as st
-
-st.set_page_config(page_title='OCR Comparator', layout ="wide")
-st.title("OCR solutions comparator")
-
-st.write("")
-st.write("")
-st.write("")
-
-st.markdown("#####  This app allows you to compare, from a given picture, the results of different solutions:")
-st.markdown("##### *EasyOcr, PaddleOCR, MMOCR, Tesseract*")
-st.write("")
-st.write("")
-
-st.markdown(''' The 1st step is to choose the language for the text recognition (not all solutions \
-support the same languages), and then choose the picture to consider. It is possible to upload a file, \
-to take a picture, or to use a demo file. \
-It is then possible to change the default values for the text area detection process, \
-before launching the detection task for each solution.''')
-st.write("")
-
-st.markdown(''' The different results are then presented. The 2nd step is to choose one of these \
-detection results, in order to carry out the text recognition process there. It is also possible to change \
-the default settings for each solution.''')
-st.write("")
-
-st.markdown("###### The recognition results appear in 2 formats:")
-st.markdown(''' - a visual format resumes the initial image, replacing the detected areas with \
-the recognized text. The background is + or - strongly colored in green according to the \
-confidence level of the recognition.
-    A slider allows you to change the font size, another \
-allows you to modify the confidence threshold above which the text color changes: if it is at \
-70% for example, then all the texts with a confidence threshold higher or equal to 70 will appear \
-in white, in black otherwise.''')
-
-st.markdown(" - a detailed format presents the results in a table, for each text box detected. \
-It is possible to download this results in a local csv file.")

requirements.txt

@@ -13,4 +13,5 @@ paddlepaddle==2.3.2
 mycolorpy==1.5.1
 plotly==5.10.0
 plotly-express==0.4.1
-pytesseract==0.3.10
+pytesseract==0.3.10
+streamlit_option_menu