Spaces:
Runtime error
Runtime error
import streamlit as st | |
from tensorflow.keras.models import load_model | |
from PIL import Image | |
import numpy as np | |
model = load_model("model_malaria_detection.h5") | |
def process_image(img): | |
img = img.convert("RGB") | |
img = img.resize((50,50)) | |
img = np.array(img) | |
if img.ndim == 2: | |
img = np.stack((img,)*3, axis=-1) | |
img = img/255.0 | |
img = np.expand_dims(img, axis=0) | |
return img | |
st.title("MALARIA RECOGNITION") | |
st.divider() | |
col1, col2, col3 = st.columns([1,2,1]) | |
with col2: | |
st.image("malaria.jpeg") | |
st.divider() | |
st.success("Upload your malaria image from blood cell and classify the images with the following labels: Uninfected and Parasitized with CNN deep learning.") | |
st.divider() | |
st.write("Upload your image and see the results") | |
st.divider() | |
file = st.file_uploader("Choose an image", type=["jpg", "jpeg", "png", "webp"]) | |
if file is not None: | |
img = Image.open(file) | |
st.image(img, caption="Uploaded image") | |
image = process_image(img) | |
prediction = model.predict(image) | |
predicted_class = np.round(prediction) | |
predicted_class = int(predicted_class.flatten()) | |
class_names = {0:"Parasitized", 1:"Uninfected"} | |
st.write(f"Predicted Malaria Type: {class_names[predicted_class]}") |