app.py

import streamlit as st
import keras.utils as image
import numpy as np
import io
import tensorflow as tf
import base64

# Load TFLite model and allocate tensors
interpreter = tf.lite.Interpreter(model_path="model/model_float16_quant.tflite")
interpreter.allocate_tensors()

st.title("Image Classifier")

uploaded_file = st.file_uploader("Upload an image...", type=["jpg", "png", "jpeg"])
img_data = None

if uploaded_file is not None:
    # Use in-memory image
    image_stream = io.BytesIO(uploaded_file.read())
    img = image.load_img(image_stream, target_size=(224, 224))

    # Convert PIL Image to data URL
    img_buffer = io.BytesIO()
    img.save(img_buffer, format="PNG")
    img_data = base64.b64encode(img_buffer.getvalue()).decode()

    # Preprocess the image
    img_array = image.img_to_array(img)
    x_mean = img_array.mean()
    x_std = img_array.std()
    img_array = (img_array - x_mean) / x_std
    img_array = np.expand_dims(img_array, axis=0)

    # Set input tensor for the TFLite interpreter
    input_details = interpreter.get_input_details()
    interpreter.set_tensor(input_details[0]['index'], img_array)

    # Invoke the interpreter
    interpreter.invoke()

    # Get the prediction result
    output_details = interpreter.get_output_details()
    predictions = interpreter.get_tensor(output_details[0]['index'])
    predicted_class = np.argmax(predictions, axis=1)

    class_dict = {
        0: 'Actinic keratoses',
        1: 'Basal cell carcinoma',
        2: 'Benign keratosis-like lesions',
        3: 'Dermatofibroma_df',
        4: 'Melanocytic nevi',
        5: 'Vascular lesions',
        6: 'Dermatofibroma_mel'
    }

    result = class_dict[predicted_class[0]]

    st.write(f"Prediction: {result}")

    # Display the uploaded image
    st.image(img, caption='Uploaded Image', use_column_width=True)