app.py

import streamlit as st
import wandb
from utils import preprocess_image, draw_boxes_on_image
import logging
import torch
import tempfile
from PIL import Image
import os
import io


# Set up logging
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)

# Initialize Weights & Biases
wandb.login(key="4462a61f98f7ea72e8dfa5f96ff9798cc1c08c60")
run = wandb.init(project="counting-trees", job_type="inference")

# Download the model
artifact = run.use_artifact('ahiamadzormaxwell7/counting-trees/model:v64', type='model')
artifact_dir = artifact.download()
logger.info(f"Model downloaded to directory: {artifact_dir}")

# Set page to wide mode
st.set_page_config(layout="wide")

# Streamlit UI
st.title("Detect Trees")
uploaded_file = st.file_uploader("Upload Aerial Image")

if uploaded_file is not None:
    # Add captions
    st.markdown("""
        <style>
        .big-font {
            font-size:20px !important;
        }
        </style>
        """, unsafe_allow_html=True)
    
    st.markdown('<p class="big-font">1 - Palm Tree (displayed in <span style="color:red">red</span>)<br>2 - Tree (displayed in <span style="color:blue">blue</span>)</p>', unsafe_allow_html=True)
    col1, col2 = st.columns(2)

    # Create a temporary file
    file_extension = os.path.splitext(uploaded_file.name)[1]
    with tempfile.NamedTemporaryFile(delete=False, suffix=file_extension) as tmp_file:
        tmp_file.write(uploaded_file.getvalue())
        tmp_file_path = tmp_file.name
    
    # Open and display original image
    image = Image.open(uploaded_file)
    col1.image(image, caption="Original Image",  width=600)
    
    # Preprocess image
    img = preprocess_image(tmp_file_path)
    logger.info("Image preprocessed.")
    
    # Load model and run prediction
    model_path = os.path.join(artifact_dir, "best_model.pth")  
    model = torch.load(model_path, map_location=torch.device('cpu'))
    model.eval()  
    
    with torch.no_grad():
        predictions = model(img)
    
    logger.info("Trees detection done.")
    
    # Process predictions
    preds = predictions[0]  
    
    # Draw boxes on image
    image_bytes = draw_boxes_on_image(img[0], preds)  
    col2.image(image_bytes, caption="Detected Trees",  width=600)

    os.unlink(tmp_file_path)

wandb.finish()