app.py

import numpy as np
import cv2
import os
from PIL import Image
import torchvision.transforms as transforms
import gradio as gr
from yolov5 import xai_yolov5
from yolov8 import xai_yolov8s

sample_images = {
    "Sample 1": os.path.join(os.getcwd(), "data/xai/sample1.jpeg"),
    "Sample 2":  os.path.join(os.getcwd(), "data/xai/sample2.jpg"),
}

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

default_sample_image = load_sample_image("Sample 1")

def load_sample_image(choice):
    if choice in sample_images:
        image_path = sample_images[choice]
        return cv2.imread(image_path)[:, :, ::-1]  
    else:
        raise ValueError("Invalid sample selection.")

def process_image(sample_choice, uploaded_image, yolo_versions=["yolov5"]):
    print(sample_choice, upload_image)
    if uploaded_image is not None:
        image = uploaded_image  # Use the uploaded image
    else:
        # Otherwise, use the selected sample image
        image = load_sample_image(sample_choice)
    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 yet implemented."))
    return result_images

with gr.Blocks() as interface:
    # Update CSS to make text light grey
    gr.HTML("""
    <style>
        body {
            background-color: black;
            color: #D3D3D3;  /* Set the default text color to light grey */
        }
        .gradio-container {
            color: #D3D3D3;  /* Ensure Gradio components also have light grey text */
        }
        h1, h2, h3, h4, h5, h6, p, label {
            color: #D3D3D3; /* Make all headings and labels light grey */
        }
        .gr-markdown {
            color: #D3D3D3; /* Ensure Markdown text is light grey */
        }
        .gr-button {
            background-color: #007bff; /* Optional: Change button background color */
            color: #D3D3D3; /* Ensure button text is light grey */
        }
        .gr-button:hover {
            background-color: #0056b3; /* Optional: Change button hover color */
        }
    </style>   
    """)
    
    gr.Markdown("<h1>XAI: Visualize Object Detection of Your Models</h1>")
    gr.Markdown("<p>Select a sample image to visualize object detection.</p>")
    
    default_sample = "Sample 1"
    
    with gr.Row(elem_classes="orchid-green-bg"):
        # Left side: Sample selection and upload image
        with gr.Column():
            sample_selection = gr.Radio(
                choices=list(sample_images.keys()),
                label="Select a Sample Image",
                type="value",
                value=default_sample,  # Set default selection
            )
            # Upload image below sample selection
            gr.Markdown("**Or upload your own image:**")
            upload_image = gr.Image(
                label="Upload an Image",
                type="pil",  # Correct type for file path compatibility
            )
        
        # Right side: Selected sample image display
        sample_display = gr.Image(
            value=load_sample_image(default_sample),  
            label="Selected Sample Image",
        )
    
    sample_selection.change(
        fn=load_sample_image,
        inputs=sample_selection,
        outputs=sample_display,
    )

    selected_models = gr.CheckboxGroup(
        choices=["yolov5", "yolov8s"],
        value=["yolov5"],
        label="Select Model(s)",
    )
    
    result_gallery = gr.Gallery(label="Results", elem_id="gallery", rows=2, height=500)

    gr.Button("Run").click(
        fn=process_image,
        inputs=[sample_selection, upload_image, selected_models],  # Include both options
        outputs=result_gallery,
    )

interface.launch(share=True)