Skip to content

Commit

Permalink
update
Browse files Browse the repository at this point in the history
  • Loading branch information
@jagadeshchilla
jagadeshchilla committed Oct 17, 2024
1 parent c376d33 commit a5e9ed8
Show file tree
Hide file tree
Showing 3 changed files with 159 additions and 142 deletions.
296 changes: 154 additions & 142 deletions pages/Image_prediction.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -13,166 +13,166 @@
import gdown
import tempfile

# Define the model links
model_links = {
'CNN': {
'url': 'https://drive.google.com/uc?id=1mtDtPtM-E7y20LlFlEPn1UI20fykFL2P',
'target_size': (128, 128)
},
'ResNet50': {
'url': 'https://drive.google.com/uc?id=1F47j2nr5JSa09mBWtUsWOkQYl5pLixqG',
'target_size': (260, 260)
},
'EfficientNet': {
'url': 'https://drive.google.com/uc?id=10mXZQWQ1RyGx6BqEsaJdcv8NcSKv-v8A',
'target_size': (260, 260)
},
'DenseNet': {
'url': 'https://drive.google.com/uc?id=14-7XGitYTJTYAksSI-LPS-b_lW7Dn8aC',
'target_size': (224, 224)
},
'VGG19': {
'url': 'https://drive.google.com/uc?id=19o-JaeGBDXpITObAVkII2sNFYp3qmGoH',
'target_size': (224, 224)
},
}

# Initialize session state
if 'saved_predictions' not in st.session_state:
st.session_state.saved_predictions = []
if 'predictions' not in st.session_state:
st.session_state.predictions = []
if 'uploaded_images' not in st.session_state:
st.session_state.uploaded_images = []
if 'model_temp_file' not in st.session_state:
st.session_state.model_temp_file = None

def load_existing_predictions():
"""Loads existing predictions from a file."""
if os.path.exists('prediction_history.json'):
with open('prediction_history.json', 'r') as f:
return json.load(f)
return []

# Load existing predictions into session state
st.session_state.saved_predictions = load_existing_predictions()

def save_predictions_to_history(uploaded_files, predictions, model_name):
"""Save predictions to a JSON file for history tracking."""
prediction_data = []
for i, uploaded_file in enumerate(uploaded_files):
actual = 'Cancer' if predictions[i][0] == 0 else 'Non Cancer'
prediction_data.append({
'file_name': uploaded_file.name,
'model_used': model_name,
'prediction': actual
})

st.session_state.saved_predictions.extend(prediction_data)

with open('prediction_history.json', 'w') as f:
json.dump(st.session_state.saved_predictions, f, indent=4)
st.success("Predictions saved to history successfully.")

cancer_warning_messages = [
"Please consult a doctor immediately.",
"We recommend scheduling a medical check-up soon.",
"It's crucial to seek medical advice right away.",
"Contact your healthcare provider for further examination.",
"This result may be concerning. Please consult a specialist."
]

def download_and_load_model(model_url):
"""Downloads and loads the model from the provided Google Drive URL."""
if st.session_state.model_temp_file is None:
with tempfile.NamedTemporaryFile(suffix='.keras', delete=False) as tmp:
st.session_state.model_temp_file = tmp.name
st.toast("📥 Downloading model... Please wait.")
with st.spinner("Downloading the model..."):
gdown.download(model_url, st.session_state.model_temp_file, quiet=False)
st.toast("✅ Model download completed!")

model = load_model(st.session_state.model_temp_file)
return model

def load_uploaded_images(uploaded_files, target_size):
"""Loads and processes uploaded images."""
images = []
for uploaded_file in uploaded_files:
image = load_img(uploaded_file, target_size=target_size)
image_array = img_to_array(image)
images.append(image_array)
return np.array(images)

def evaluate_model(model, images):
"""Evaluate the model on the uploaded images and return predictions."""
predictions = model.predict(images)
predicted_classes = (predictions > 0.5).astype(int)
return predicted_classes

def display_predictions(uploaded_files, predictions):
"""Display the predictions along with the images."""
for i, uploaded_file in enumerate(uploaded_files):
actual = 'Cancer' if predictions[i][0] == 0 else 'Non Cancer'
caption = f'Predicted: {actual}'
st.image(uploaded_file, caption=caption, use_column_width=True)

if actual == 'Cancer':
warning_message = random.choice(cancer_warning_messages)
st.warning(warning_message)

def handle_prediction_and_display(model_url, uploaded_files, target_size):
"""Handle prediction and display results for uploaded images."""
st.info("Downloading and loading the model. This may take a few moments...")
with st.spinner("Loading model..."):
model = download_and_load_model(model_url)

images = load_uploaded_images(uploaded_files, target_size)
predictions = evaluate_model(model, images)
st.session_state.predictions = predictions
st.session_state.uploaded_images = uploaded_files

st.toast("✨ Images predicted successfully!")
st.subheader('Predictions:')
display_predictions(uploaded_files, predictions)

def clear_predictions():
"""Clear all session state related to predictions and uploaded images."""
st.session_state.predictions = []
st.session_state.uploaded_images = []
st.session_state.model_temp_file = None
st.success("🗑️ Cleared all predictions and uploaded images.")

def image_to_base64(image: Image.Image) -> str:
"""Convert an image to base64 for displaying on Streamlit."""
buffered = io.BytesIO()
image.save(buffered, format="PNG")
return base64.b64encode(buffered.getvalue()).decode()

def show_image_prediction():
# Define the model links
model_links = {
'CNN': {
'url': 'https://drive.google.com/uc?id=1mtDtPtM-E7y20LlFlEPn1UI20fykFL2P',
'target_size': (128, 128)
},
'ResNet50': {
'url': 'https://drive.google.com/uc?id=1F47j2nr5JSa09mBWtUsWOkQYl5pLixqG',
'target_size': (260, 260)
},
'EfficientNet': {
'url': 'https://drive.google.com/uc?id=10mXZQWQ1RyGx6BqEsaJdcv8NcSKv-v8A',
'target_size': (260, 260)
},
'DenseNet': {
'url': 'https://drive.google.com/uc?id=14-7XGitYTJTYAksSI-LPS-b_lW7Dn8aC',
'target_size': (224, 224)
},
'VGG19': {
'url': 'https://drive.google.com/uc?id=19o-JaeGBDXpITObAVkII2sNFYp3qmGoH',
'target_size': (224, 224)
},
}

# Initialize session state
if 'saved_predictions' not in st.session_state:
st.session_state.saved_predictions = []
if 'predictions' not in st.session_state:
st.session_state.predictions = []
if 'uploaded_images' not in st.session_state:
st.session_state.uploaded_images = []
if 'model_temp_file' not in st.session_state:
st.session_state.model_temp_file = None

def load_existing_predictions():
if os.path.exists('prediction_history.json'):
with open('prediction_history.json', 'r') as f:
return json.load(f)
return []

# Load existing predictions into session state
st.session_state.saved_predictions = load_existing_predictions()

def save_predictions_to_history(uploaded_files, predictions, model_name):
prediction_data = []
for i, uploaded_file in enumerate(uploaded_files):
actual = 'Cancer' if predictions[i][0] == 0 else 'Non Cancer'
prediction_data.append({
'file_name': uploaded_file.name,
'model_used': model_name,
'prediction': actual
})

st.session_state.saved_predictions.extend(prediction_data)

with open('prediction_history.json', 'w') as f:
json.dump(st.session_state.saved_predictions, f, indent=4)
st.success("Predictions saved to history successfully.")

cancer_warning_messages = [
"Please consult a doctor immediately.",
"We recommend scheduling a medical check-up soon.",
"It's crucial to seek medical advice right away.",
"Contact your healthcare provider for further examination.",
"This result may be concerning. Please consult a specialist."
]

def download_and_load_model(model_url):
"""Downloads and loads the model from the provided Google Drive URL."""
if st.session_state.model_temp_file is None:
with tempfile.NamedTemporaryFile(suffix='.keras', delete=False) as tmp:
st.session_state.model_temp_file = tmp.name
# Show toast message for downloading model
st.toast("📥 Downloading model... Please wait.")
# Spinner for downloading the model
with st.spinner("Downloading the model..."):
gdown.download(model_url, st.session_state.model_temp_file, quiet=False)

# Show toast message for download completion
st.toast("✅ Model download completed!")

# Load the model from the temp file
model = load_model(st.session_state.model_temp_file)
return model

# Streamlit UI
st.title('Oral Cancer Detection Model Evaluation')

# Model selection
model_selection = st.selectbox("Select a model", list(model_links.keys()))
uploaded_files = st.file_uploader("Upload images", type=['jpg', 'jpeg', 'png'], accept_multiple_files=True)

# Upload images
uploaded_files = st.file_uploader(
"Upload images", type=['jpg', 'jpeg', 'png'], accept_multiple_files=True)

if uploaded_files:
target_size = model_links[model_selection]['target_size']

def load_uploaded_images(uploaded_files, target_size):
images = []
for uploaded_file in uploaded_files:
image = load_img(uploaded_file, target_size=target_size)
image_array = img_to_array(image)
images.append(image_array)
return np.array(images)

X_test = load_uploaded_images(uploaded_files, target_size)

# Function to evaluate the model on uploaded images
def evaluate_model(model, images):
predictions = model.predict(images)
predicted_classes = (predictions > 0.5).astype(int)
return predicted_classes

# Add a button to trigger predictions
if st.button('Predict'):
handle_prediction_and_display(model_links[model_selection]['url'], uploaded_files, target_size)
st.info("Downloading and loading the model. This may take a few moments...")

# Download and load the model
model_url = model_links[model_selection]['url']
with st.spinner("Loading model..."):
model_to_use = download_and_load_model(model_url)

# Evaluate the model
with st.spinner("Evaluating images..."):
st.session_state.predictions = evaluate_model(model_to_use, X_test)
st.session_state.uploaded_images = uploaded_files

# Show toast message for image prediction
st.toast("✨ Images predicted successfully!")

# Display predictions
st.subheader('Predictions:')
for i, uploaded_file in enumerate(uploaded_files):
actual = 'Cancer' if st.session_state.predictions[i][0] == 0 else 'Non Cancer'
caption = f'Predicted: {actual}'
st.image(uploaded_file, caption=caption, use_column_width=True)

if actual == 'Cancer':
warning_message = random.choice(cancer_warning_messages)
st.warning(warning_message)

col1, col2 = st.columns(2)

with col1:
if st.button('Clear'):
clear_predictions()
st.session_state.predictions = []
st.session_state.uploaded_images = []
st.session_state.model_temp_file = None # Reset the temp file
st.success("🗑️ Cleared all predictions and uploaded images.")

with col2:
if st.session_state.predictions and st.session_state.uploaded_images:
if len(st.session_state.predictions) > 0 and len(st.session_state.uploaded_images) > 0:
if st.button('Save Predictions'):
save_predictions_to_history(st.session_state.uploaded_images, st.session_state.predictions, model_selection)
save_predictions_to_history(
st.session_state.uploaded_images, st.session_state.predictions, model_selection)

# Download predictions functionality
if st.session_state.predictions and st.session_state.uploaded_images:
if len(st.session_state.predictions) > 0 and len(st.session_state.uploaded_images) > 0:
prediction_images = []
for i, uploaded_file in enumerate(st.session_state.uploaded_images):
actual = 'Cancer' if st.session_state.predictions[i][0] == 0 else 'Non Cancer'
Expand All @@ -187,6 +187,7 @@ def show_image_prediction():
buf.seek(0)
prediction_images.append((buf, f'prediction_{i + 1}.png'))

# Create zip file for download
zip_buffer = io.BytesIO()
with zipfile.ZipFile(zip_buffer, 'w') as zf:
for image_buf, filename in prediction_images:
Expand All @@ -200,16 +201,27 @@ def show_image_prediction():
mime='application/zip'
)

# Utility function to convert an image to base64 for display
def image_to_base64(image: Image.Image) -> str:
buffered = io.BytesIO()
image.save(buffered, format="PNG")
return base64.b64encode(buffered.getvalue()).decode()

# Display logo
logo_path = "./assets/logo.png" # Update with your logo file path
logo_image = Image.open(logo_path)

# Convert the logo image to base64
logo_base64 = image_to_base64(logo_image)

# Display the logo with custom CSS styles
st.sidebar.markdown(
f"""
<img src="data:image/jpeg;base64,{logo_base64}"
style="border-radius: 30px; box-shadow: 0 4px 8px rgba(0, 0, 0, 0.2); width: 90%; height: auto;" />
""", unsafe_allow_html=True
)

# Call the function to show image prediction
if __name__ == "__main__":
show_image_prediction()
Binary file modified pages/__pycache__/Image_prediction.cpython-312.pyc
View file
Binary file not shown.
5 changes: 5 additions & 0 deletions prediction_history.json
View file
Original file line number Diff line number Diff line change
Expand Up @@ -163,5 +163,10 @@
"file_name": "008.jpeg",
"model_used": "CNN",
"prediction": "Cancer"
},
{
"file_name": "006.jpeg",
"model_used": "CNN",
"prediction": "Cancer"
}
]

0 comments on commit a5e9ed8

Please sign in to comment.