detect_app.py

import torch
import warnings
import gradio as gr
import cv2
import torchvision
from torch import nn
from torchvision.models import mobilenet_v3_small
import numpy as np

device = "cuda" if torch.cuda.is_available() else "cpu"
warnings.filterwarnings("ignore")


def flip_text(x):
    return x[::-1]


def method1_prep(image):
    transforms = torchvision.models.MobileNet_V3_Small_Weights.IMAGENET1K_V1.transforms()
    image = torch.from_numpy(image).permute(2, 0, 1)
    return transforms(image)


def method2_prep(image):
    transforms = torchvision.transforms.Compose([
        torchvision.transforms.Resize((256, 256)),
        torchvision.transforms.CenterCrop((224, 224))
    ])
    t_lower = 50
    t_upper = 150
    
    height, width = image.shape[:2]

    x = (width - 1920) // 2
    y = (height - 1080) // 2
    
    image = image[y:y+1080, x:x+1920]

    img = torch.from_numpy(cv2.Canny(image, t_lower, t_upper)[np.newaxis, ...])
    img = torch.vstack((img, img, img))

    return transforms(img.type(torch.float32))


def model1_inf(x):
    print("Method 1")

    image = method1_prep(x).unsqueeze(dim=0)
    model = mobilenet_v3_small(weights='DEFAULT')
    model.classifier[3] = nn.Linear(in_features=1024, out_features=2, bias=True)

    model.load_state_dict(torch.load('./weights/method1(0.668).pt'))

    model.eval()  # Set the model to evaluation mode

    with torch.inference_mode():
        model = model.to(device)
        image = image.to(device)
        output = torch.softmax(model(image), dim=1).detach().cpu()
        prediction = torch.argmax(output, dim=1).item()
        del model
        torch.cuda.empty_cache()
        if prediction == 0:
            return "The image is not pixelated"
        else:
            return "The image is pixelated"


def model2_inf(x):
    print("Method 2")

    image = method2_prep(x).unsqueeze(dim=0)
    model = mobilenet_v3_small(weights='DEFAULT')
    model.classifier[3] = nn.Linear(in_features=1024, out_features=2, bias=True)

    image_np = image[0].permute(1, 2, 0).cpu().numpy()
    image_np = (image_np * 255).astype(np.uint8)  # Ensure the image is of type uint8

    model.load_state_dict(torch.load('./weights/method2(0.960).pt'))
    #print("\nModel weights loaded successfully")

    model.eval()  # Set the model to evaluation mode

    with torch.inference_mode():
        model = model.to(device)
        image = image.to(device)
        output = torch.softmax(model(image), dim=1).detach().cpu()
        prediction = torch.argmax(output, dim=1).item()
        del model
        torch.cuda.empty_cache()
        if prediction == 0:
            return "The image is not pixelated"
        else:
            return "The image is pixelated"


with gr.Blocks() as app:
    gr.Markdown("### Pixelation Detection App")

    # Selecting method 
    method = gr.Radio(["Method 1", "Method 2 (Proposed Method)"], label="Select Method")

    with gr.Tab("Classification by image"):
        image_input = gr.Image(type="numpy", label="Upload an image")
        output_text = gr.Textbox(label="Output")

        def classify_image(img, method):
            if method == "Method 1":
                return model1_inf(img)
            else:
                return model2_inf(img)

        method.change(fn=classify_image, inputs=[image_input, method], outputs=output_text)
        image_input.change(fn=classify_image, inputs=[image_input, method], outputs=output_text)

        
app.launch(share=False)