Upload 3 files

app.py

@@ -0,0 +1,70 @@
+import streamlit as st
+from PIL import Image
+import numpy as np
+import plotly.graph_objects as go
+from models import predict
+import cv2
+
+
+
+google_form_link = 'https://docs.google.com/forms/d/1xKeveRFf90_wCX-tjMInFC48XmFF8HOsPSQ47ruOFk0/edit'
+# Load the pre-trained Haar Cascade for eye detection
+eye_cascade = cv2.CascadeClassifier(cv2.data.haarcascades + 'haarcascade_eye.xml')
+
+# Function to check if the image contains an eye
+def contains_eye(image):
+    gray_image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
+    eyes = eye_cascade.detectMultiScale(gray_image, scaleFactor=1.1, minNeighbors=5, minSize=(30, 30))
+    return len(eyes) > 0
+
+# Function to load the image
+def load_image(image_file):
+    img = Image.open(image_file)
+    return img
+
+# Streamlit app title
+st.title("Eye Cataract Detection")
+
+# Streamlit header and subheader
+st.header('Upload the image')
+
+# File uploader widget
+image_file = st.file_uploader("Upload Images", type=["png", "jpg", "jpeg"])
+
+# Check if an image file is uploaded
+if image_file is not None:
+    img = load_image(image_file)
+
+    # Display the uploaded image
+    st.image(img, width=250)
+
+    # Convert image to OpenCV format
+    open_cv_image = np.array(img)
+
+    if contains_eye(open_cv_image):
+        # Button for detection
+        if st.button('Detect'):  # This line adds a 'Detect' button
+            # Predict the label and probability
+            label, prob = predict(open_cv_image)
+
+            # Use markdown to style the text and include emojis
+            if prob > 0.5:
+                st.markdown(f"<h2 style='color: red;'>Cataract Detected 😟</h2>", unsafe_allow_html=True)
+                #st.markdown(f"### Probability: **{prob:.2f}**")
+            else:
+                st.markdown(f"<h2 style='color: green;'>No Cataract Detected 😄</h2>", unsafe_allow_html=True)
+                #st.markdown(f"### Probability: **{prob:.2f}**")
+
+            # Pie chart visualization
+            fig = go.Figure(data=[go.Pie(labels=['Cataract', 'No Cataract'], 
+                                        values=[prob, 1 - prob],
+                                        hoverinfo='label+percent', 
+                                        pull=[0, 0])])
+            fig.update_layout(title_text='Cataract Detection Probability')
+            st.plotly_chart(fig)
+
+            st.subheader("Doctor's Verification")
+            st.markdown(f"[Click here to provide feedback on the cataract detection results]({google_form_link})", unsafe_allow_html=True)
+
+    else:
+        st.error("No eyes detected in the image. Please upload a relevant eye image.")

models.py

@@ -0,0 +1,82 @@
+import numpy as np
+import pandas as pd
+import cv2 as cv
+import streamlit as st
+import matplotlib.pyplot as plt
+from skimage.feature import graycomatrix, graycoprops
+import joblib
+
+indextable = ['dissimilarity', 'contrast',
+              'homogeneity', 'energy', 'correlation', 'Label']
+obj = {
+    0.0: "Normal",
+    1.0: "Cataract",
+    2.0: "Glaucoma",
+    3.0: 'Retina Disease'
+}
+width, height = 400, 400
+distance = 10
+teta = 90
+
+# Code to extract features from Image using Gray Level Co occurrence Image
+
+
+def get_feature(matrix, name):
+    feature = graycoprops(matrix, name)
+    result = np.average(feature)
+    return result
+
+
+def preprocessingImage(image):
+    test_img = cv.cvtColor(image, cv.COLOR_BGR2RGB)
+    test_img_gray = cv.cvtColor(test_img, cv.COLOR_RGB2GRAY)
+    test_img_thresh = cv.adaptiveThreshold(
+        test_img_gray, 255, cv.ADAPTIVE_THRESH_GAUSSIAN_C, cv.THRESH_BINARY_INV, 11, 3)
+
+    cnts = cv.findContours(
+        test_img_thresh, cv.RETR_EXTERNAL, cv.CHAIN_APPROX_SIMPLE)
+    cnts = cnts[0] if len(cnts) == 2 else cnts[1]
+    cnts = sorted(cnts, key=cv.contourArea, reverse=True)
+
+    for c in cnts:
+        x, y, w, h = cv.boundingRect(c)
+        test_img_ROI = test_img[y:y+h, x:x+w]
+        break
+
+    test_img_ROI_resize = cv.resize(test_img_ROI, (width, height))
+    test_img_ROI_resize_gray = cv.cvtColor(
+        test_img_ROI_resize, cv.COLOR_RGB2GRAY)
+
+    return test_img_ROI_resize_gray
+
+
+def extract(path):
+    data_eye = np.zeros((5, 1))
+
+    # path = cv.imread(path)
+    img = preprocessingImage(path)
+
+    glcm = graycomatrix(img, [distance], [teta],
+                        levels=256, symmetric=True, normed=True)
+
+    for i in range(len(indextable[:-1])):
+        features = []
+        feature = get_feature(glcm, indextable[i])
+        features.append(feature)
+        data_eye[i, 0] = features[0]
+    return pd.DataFrame(np.transpose(data_eye), columns=indextable[:-1])
+
+
+"""
+Return predicted class with its probability
+"""
+
+model = joblib.load("model.pkl")
+
+
+
+def predict(path):
+    X = extract(path)
+    y = model.predict(X)[0]
+    prob = model.predict_proba(X)[0, int(y)]
+    return (obj[y], prob)

requirements.txt

@@ -0,0 +1,11 @@
+numpy
+pandas
+opencv-python
+matplotlib
+scikit-learn
+scikit-image
+streamlit
+joblib
+pillow
+plotly
+opencv-python-headless==4.5.4.60