Nada2001/streamlitSegmentation

1import streamlit as st2 3import tensorflow as tf4from PIL import Image5import numpy as np6import cv27 8 9 10from huggingface_hub import from_pretrained_keras11try:12    model=from_pretrained_keras("Nada2001/streamlitSegmentation")13except:14    model=tf.keras.models.load_model('/content/drive/MyDrive/dataX-ray_modelH5/UNet data xray_model3.h5')15    pass16    17st.header("Segmentation of Teeth in Panoramic X-ray Image Using UNet")18 19examples=["1.jpg","2.jpg","3.jpg"]20 21def load_image(image_file):22	img = Image.open(image_file)23	return img24 25def convert_one_channel(img):26    #some images have 3 channels , although they are grayscale image27    if len(img.shape)>2:28        img= cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)29        return img30    else:31        return img32    33def convert_rgb(img):34    #some images have 3 channels , although they are grayscale image35    if len(img.shape)==2:36        img= cv2.cvtColor(img,cv2.COLOR_GRAY2RGB)  37        return img38    else:39        return img40    41    42st.subheader("Upload Dental Panoramic X-ray Image Image")43image_file = st.file_uploader("Upload Images", type=["png","jpg","jpeg"])44 45 46col1, col2, col3 = st.columns(3)47with col1:48    ex=load_image(examples[0])49    st.image(ex,width=200)50    if st.button('Example 1'):51        image_file=examples[0]52 53with col2:54    ex1=load_image(examples[1])55    st.image(ex1,width=200)56    if st.button('Example 2'):57        image_file=examples[1]58 59 60with col3:61    ex2=load_image(examples[2])62    st.image(ex2,width=200)63    if st.button('Example 3'):64        image_file=examples[2]65    66    67if image_file is not None:68 69      img=load_image(image_file)70      71      st.text("Making A Prediction ....")72      st.image(img,width=850)73      74      img=np.asarray(img)75  76      img_cv=convert_one_channel(img)77      img_cv=cv2.resize(img_cv,(512,512), interpolation=cv2.INTER_LANCZOS4)78      img_cv=np.float32(img_cv/255)79      80      img_cv=np.reshape(img_cv,(1,512,512,1))81      prediction=model.predict(img_cv)82      predicted=prediction[0]83      predicted = cv2.resize(predicted, (img.shape[1],img.shape[0]), interpolation=cv2.INTER_LANCZOS4)84      mask=np.uint8(predicted*255)# 85      _, mask = cv2.threshold(mask, thresh=0, maxval=255, type=cv2.THRESH_BINARY+cv2.THRESH_OTSU)86      kernel =( np.ones((5,5), dtype=np.float32))87      mask=cv2.morphologyEx(mask, cv2.MORPH_OPEN, kernel,iterations=1 )  88      mask=cv2.morphologyEx(mask, cv2.MORPH_CLOSE, kernel,iterations=1 )89      cnts,hieararch=cv2.findContours(mask,cv2.RETR_TREE,cv2.CHAIN_APPROX_SIMPLE)90      output = cv2.drawContours(convert_rgb(img), cnts, -1, (255, 0, 0) , 3)91 92 93      if output is not None :      94          st.subheader("Predicted Image")  95          st.write(output.shape)96          st.image(output,width=850)97 98      st.text("DONE ! ....")99