import streamlit as st
import cv2
import numpy as np
import torch
from torchvision import transforms

# Load YOLOv5 models
models = []
models.append(torch.hub.load('ultralytics/yolov5', 'yolov5s', pretrained=True))
models.append(torch.hub.load('ultralytics/yolov5', 'yolov5m', pretrained=True))
models.append(torch.hub.load('ultralytics/yolov5', 'yolov5l', pretrained=True))
models.append(torch.hub.load('ultralytics/yolov5', 'yolov5x', pretrained=True))

# Custom CSS
html_style = """
<style>
.container {
    padding: 20px;
    background-color: #f9f9f9;
    border-radius: 10px;
    box-shadow: 0 4px 8px rgba(0, 0, 0, 0.1);
}
.title {
    color: #ff69b4;
    font-size: 36px;
    text-align: center;
    margin-bottom: 30px;
}
.subheader {
    color: #ff69b4;
    font-size: 24px;
    margin-top: 20px;
}
.image-container {
    margin-top: 20px;
    text-align: center;
}
</style>
"""

st.markdown(html_style, unsafe_allow_html=True)

st.markdown("<h1 class='title'>AI Skin Analyzer</h1>", unsafe_allow_html=True)

uploaded_file = st.file_uploader("Choose an image...", type=["jpg", "jpeg", "png"])

if uploaded_file is not None:
    image = cv2.imdecode(np.fromstring(uploaded_file.read(), np.uint8), 1)
    st.image(image, caption="Uploaded Image", use_column_width=True)
    st.markdown("<h2 class='subheader'>Model Predictions:</h2>", unsafe_allow_html=True)
    
    # Perform object detection for each model
    for i, model in enumerate(models):
        st.markdown(f"<h3 class='subheader'>Model {i+1}</h3>", unsafe_allow_html=True)
        results = model(image)
        results.render()
        output_image = results.imgs[0]
        st.image(output_image, use_column_width=True)