app.py

import gradio as gr
import netron
import os
import threading
import time    
from PIL import Image
import cv2
import numpy as np
from yolov5 import xai_yolov5
from yolov8 import xai_yolov8s
import requests
"""
# Sample images directory
sample_images = {
    "Sample 1": os.path.join(os.getcwd(), "data/xai/sample1.jpeg"),
    "Sample 2": os.path.join(os.getcwd(), "data/xai/sample2.jpg"),
}

# Preloaded model file path
preloaded_model_file = os.path.join(os.getcwd(), "weight_files/yolov5.onnx")  # Example path

# Function to load sample image
def load_sample_image(sample_name):
    image_path = sample_images.get(sample_name)
    if image_path and os.path.exists(image_path):
        return Image.open(image_path)
    return None

# Function to process the image
def process_image(sample_choice, uploaded_image, yolo_versions):
    # Use uploaded or sample image
    if uploaded_image is not None:
        image = uploaded_image
    else:
        image = load_sample_image(sample_choice)

    # Resize and process the image
    image = np.array(image)
    image = cv2.resize(image, (640, 640))
    result_images = []

    for yolo_version in yolo_versions:
        if yolo_version == "yolov5":
            result_images.append(xai_yolov5(image))
        elif yolo_version == "yolov8s":
            result_images.append(xai_yolov8s(image))
        else:
            result_images.append((Image.fromarray(image), f"{yolo_version} not implemented."))

    return result_images

# Start Netron backend
def start_netron_backend(model_file):
    def serve_netron():
        netron.start(model_file, address=("0.0.0.0", 8080), browse=False)
        #netron.start(model_file, address="0.0.0.0:8080", browse=False)  # Updated Netron arguments

    # Launch Netron in a separate thread
    threading.Thread(target=serve_netron, daemon=True).start()

    # Wait until Netron server is ready
    def wait_for_netron(url, timeout=10):
        start_time = time.time()
        while time.time() - start_time < timeout:
            try:
                response = requests.get(url)
                if response.status_code == 200:
                    return True
            except requests.ConnectionError:
                time.sleep(0.5)
        return False

    # Check server readiness
    wait_for_netron("http://localhost:8080/", timeout=15)

# View Netron model
def view_netron_model():
    # Ensure model exists
    if not os.path.exists(preloaded_model_file):
        return "Model file not found."

    # Start Netron backend
    start_netron_backend(preloaded_model_file)
    return gr.HTML('<iframe src="http://localhost:8080/" width="100%" height="600px"></iframe>')

# Custom CSS for styling (optional)
custom_css = """
#run_button {
   # background-color: purple;
   # color: white;
  #  width: 120px;
  #  border-radius: 5px;
  #  font-size: 14px;
#}
"""

# Gradio UI
with gr.Blocks(css=custom_css) as interface:
    gr.Markdown("# XAI: Visualize Object Detection of Your Models")

    # Default sample
    default_sample = "Sample 1"

    with gr.Row():
        # Left: Select sample or upload image
        with gr.Column():
            sample_selection = gr.Radio(
                choices=list(sample_images.keys()),
                label="Select a Sample Image",
                type="value",
                value=default_sample,
            )

            upload_image = gr.Image(label="Upload an Image", type="pil")

            selected_models = gr.CheckboxGroup(
                choices=["yolov5", "yolov8s"],
                value=["yolov5"],
                label="Select Model(s)",
            )

            run_button = gr.Button("Run", elem_id="run_button")

        # Right: Display sample image
        with gr.Column():
            sample_display = gr.Image(
                value=load_sample_image(default_sample),
                label="Selected Sample Image",
            )

    # Results and Netron
    with gr.Row():
        result_gallery = gr.Gallery(
            label="Results",
            elem_id="gallery",
            rows=1,
            height=500,
        )

        # Display Netron iframe
        netron_display = gr.HTML(view_netron_model())

    # Sample selection update
    sample_selection.change(
        fn=load_sample_image,
        inputs=sample_selection,
        outputs=sample_display,
    )

    # Process image
    run_button.click(
        fn=process_image,
        inputs=[sample_selection, upload_image, selected_models],
        outputs=[result_gallery],
    )

# Launch Gradio app
if __name__ == "__main__":
    interface.launch(share=True)
"""

import gradio as gr
import torch
from ultralytics import YOLO
import os

# Assuming the YOLOv5 model is located in the same directory as this script
model_path = os.path.join(os.getcwd(), "weight_files/yolov5.onnx")   # Replace with the actual model directory name

def visualize_yolov5():
  """
  Visualizes the given YOLOv5 model using Netron.
  """
  try:
    # Load the YOLOv5 model
    model = YOLO(model_path) 

    # Extract the PyTorch model 
    pytorch_model = model.model

    # Save the PyTorch model to a temporary file
    temp_model_path = "temp_model" 
    torch.save(pytorch_model.state_dict(), os.path.join(temp_model_path, "pytorch_model.bin"))

    # Run Netron 
    os.system(f"netron {temp_model_path}") 

  except Exception as e:
    print(f"Error visualizing model: {e}")

# Create the Gradio interface
iface = gr.Interface(
    fn=visualize_yolov5,
    inputs=None,  # No input required
    outputs="text", 
    title="Netron YOLOv5 Model Visualization",
    description="Visualize the YOLOv5 model."
)

# Launch the Gradio app
iface.launch(share=True)