app.py

import gradio as gr
import numpy as np
from PIL import Image, ImageDraw
import torch
from torchvision import transforms
from transformers import AutoModelForImageClassification, AutoFeatureExtractor

# Define all available models
MODEL_LIST = {
    'beit': "microsoft/beit-base-patch16-224-pt22k-ft22k",
    'vit': "google/vit-base-patch16-224",
    'convnext': "facebook/convnext-tiny-224",
}

# Preprocessing transforms
def get_preprocessor(model_name):
    extractor = AutoFeatureExtractor.from_pretrained(MODEL_LIST[model_name])
    return extractor

# Load a model from Hugging Face
def load_model(model_name):
    model = AutoModelForImageClassification.from_pretrained(MODEL_LIST[model_name]).cuda().eval()
    return model

# Function to make predictions
def predict(image, model, preprocessor):
    inputs = preprocessor(images=image, return_tensors="pt").to("cuda")
    with torch.no_grad():
        outputs = model(**inputs)
        predicted_class = torch.argmax(outputs.logits, dim=1).item()
    return model.config.id2label[predicted_class]

# Function to draw a rectangle on the image
def draw_rectangle(image, x, y, size=224):
    image_pil = image.copy()  # Create a copy to avoid modifying the original image
    draw = ImageDraw.Draw(image_pil)
    x1, y1 = x, y
    x2, y2 = x + size, y + size
    draw.rectangle([x1, y1, x2, y2], outline="red", width=5)
    return image_pil

# Function to crop the image
def crop_image(image, x, y, size=224):
    image_np = np.array(image)
    h, w, _ = image_np.shape
    x = min(max(x, 0), w - size)
    y = min(max(y, 0), h - size)
    cropped = image_np[y:y+size, x:x+size]
    return Image.fromarray(cropped)

# Global variables
current_model = None
current_preprocessor = None

# Gradio Interface
with gr.Blocks() as demo:
    gr.Markdown("## Test Public Models for Coral Classification")

    with gr.Row():
        with gr.Column():
            model_selector = gr.Dropdown(choices=list(MODEL_LIST.keys()), value='beit', label="Select Model")
            image_input = gr.Image(type="pil", label="Upload Image", interactive=True)
            x_slider = gr.Slider(minimum=0, maximum=1000, step=1, value=0, label="X Coordinate")
            y_slider = gr.Slider(minimum=0, maximum=1000, step=1, value=0, label="Y Coordinate")
        with gr.Column():
            interactive_image = gr.Image(label="Interactive Image with Selection")
            cropped_image = gr.Image(label="Cropped Patch")
            label_output = gr.Textbox(label="Predicted Label")

    # Update the current model and preprocessor
    def update_model(model_name):
        global current_model, current_preprocessor
        current_model = load_model(model_name)
        current_preprocessor = get_preprocessor(model_name)
        return f"Model {model_name} loaded successfully."

    # Update the rectangle and crop the patch
    def update_selection(image, x, y):
        overlay_image = draw_rectangle(image, x, y)
        cropped = crop_image(image, x, y)
        return overlay_image, cropped

    # Predict the label from the cropped patch
    def predict_from_cropped(cropped):
        return predict(cropped, current_model, current_preprocessor)

    # Buttons and interactions
    crop_button = gr.Button("Crop")
    crop_button.click(fn=update_selection, inputs=[image_input, x_slider, y_slider], outputs=[interactive_image, cropped_image])

    predict_button = gr.Button("Predict")
    predict_button.click(fn=predict_from_cropped, inputs=cropped_image, outputs=label_output)

    model_selector.change(fn=update_model, inputs=model_selector, outputs=None)

    # Update sliders dynamically based on uploaded image size
    def update_sliders(image):
        if image is not None:
            width, height = image.size
            return gr.update(maximum=width - 224), gr.update(maximum=height - 224)
        return gr.update(), gr.update()

    image_input.change(fn=update_sliders, inputs=image_input, outputs=[x_slider, y_slider])

demo.launch(server_name="0.0.0.0", server_port=7860)