app.py

import torch, torchvision
from torchvision import transforms
import numpy as np
import gradio as gr
from PIL import Image
from pytorch_grad_cam import GradCAM
from pytorch_grad_cam.utils.image import show_cam_on_image
import gradio as gr
import os

from helper import CifarAlbumentations, get_train_transforms, get_test_transforms
from resnet import CustomResNet

config = {
    'batch_size': 512,
    'data_dir': './data',
    'classes': ['airplane', 'automobile', 'bird', 'cat', 'deer', 'dog', 'frog', 'horse', 'ship', 'truck'],
    'num_classes': 10,
    'lr': 0.01,
    'max_lr': 0.1,
    'max_lr_epoch': 5,
    'dropout' : 0.01,
    'LEARNING_RATE' : 1e-5,
    'WEIGHT_DECAY' : 1e-4,
    'NUM_EPOCHS' : 100
}

train_transforms = get_train_transforms()
test_transforms = get_test_transforms()
model = CustomResNet(config, config['dropout'], train_transforms, test_transforms)
model.load_state_dict(torch.load("resnet_model_v7.pth", map_location=torch.device('cpu')), strict=False)
model.setup(stage="test")

inv_normalize = transforms.Normalize(
    mean=[-0.50/0.23, -0.50/0.23, -0.50/0.23],
    std=[1/0.23, 1/0.23, 1/0.23]
)

classes = ('plane', 'car', 'bird', 'cat', 'deer',
           'dog', 'frog', 'horse', 'ship', 'truck')

classes_for_categorize = {0: 'plane',
           1: 'car',
           2: 'bird',
           3: 'cat',
           4: 'deer',
           5: 'dog',
           6: 'frog',
           7: 'horse',
           8: 'ship',
           9: 'truck'}


def inference(input_img, transparency=0.5, target_layer_number=-1, top_classes=10):
    transform = transforms.ToTensor()
    org_img = input_img
    input_img = transform(input_img)
    input_img = input_img
    input_img = input_img.unsqueeze(0)
    outputs = model(input_img)
    softmax = torch.nn.Softmax(dim=0)
    o = softmax(outputs.flatten())
    confidences = {classes[i]: float(o[i]) for i in range(10)}
    sorted_classes = sorted(confidences.items(), key=lambda x: x[1], reverse=True)
    top_classes = sorted_classes[:top_classes]
    top_classes_dict = {cls: conf for cls, conf in top_classes}

    _, prediction = torch.max(outputs, 1)
    target_layers = [model.conv2]
    cam = GradCAM(model=model, target_layers=target_layers, use_cuda=False)
    grayscale_cam = cam(input_tensor=input_img, targets=None)
    grayscale_cam = grayscale_cam[0, :]
    img = input_img.squeeze(0)
    img = inv_normalize(img)
    rgb_img = np.transpose(img, (1, 2, 0))
    rgb_img = rgb_img.numpy()
    visualization = show_cam_on_image(org_img/255, grayscale_cam, use_rgb=True, image_weight=transparency)
    return top_classes_dict, visualization

def show_misclassified_images_wrap(num_images=10, use_gradcam=False, gradcam_layer=-1, transparency=0.5):
    transparency = float(transparency)
    num_images = int(num_images)
    if use_gradcam == "Yes":
      use_gradcam = True
    else:
      use_gradcam = False

    return model.show_misclassified_images(num_images, use_gradcam, gradcam_layer, transparency)

title = "CIFAR10 Image Classification"
description = "Upload an Image or Choose from Examples Below"

images_folder = "examples"
# Define the examples list with full paths
examples = [[os.path.join(images_folder, "plane.jpg"), 0.5, -1,10],
            [os.path.join(images_folder, "car.jpg"), 0.5, -1,5],
            [os.path.join(images_folder, "bird.jpg"), 0.5, -1,3],
            [os.path.join(images_folder, "cat.jpg"), 0.5, -1, 5],
            [os.path.join(images_folder, "deer.jpg"), 0.5, -1,7],
            [os.path.join(images_folder, "dog.jpg"), 0.5, -1,6],
            [os.path.join(images_folder, "frog.jpg"), 0.5, -1,2],
            [os.path.join(images_folder, "horse.jpg"), 0.5, -1,10],
            [os.path.join(images_folder, "ship.jpg"), 0.5, -1,10],
            [os.path.join(images_folder, "truck.jpeg"), 0.5, -1,10]]

# Create the input interface with the modified template
input_interface = gr.Interface(
    inference,
    inputs=[
        gr.Image(shape=(32, 32), label="Input Image"),
        gr.Slider(0, 1, value=0.5, label="Transparency", info="Set the Opacity of CAM"),
        gr.Slider(-2, -1, value=-2, step=1, label="Network Layer", info="GradCAM Network Layer"),
        gr.Slider(1, 10, step=1, value=10, label="Top Classes", info="How many top classes do you want to view")
    ],
    outputs=[
        gr.Label(num_top_classes=10),
        gr.Image(shape=(32, 32), label="Model Prediction").style(width=300, height=300)
    ],
    description=description,
    examples=[[f'examples/{k}.jpg'] for k in classes_for_categorize.values()],)

mislclassified_description = "Misclassified Image for Custom Resnet"
icon_html = '<i class="fas fa-chart-bar"></i>'
title_with_icon = f"""
<div style="background-color: #f1f4f0; padding: 10px; display: flex; align-items: center;">
    {icon_html} <span style="margin-left: 10px;">Custom Resnet on CIFAR10 using PyTorch Lightning and GradCAM</span>
</div>
"""
# Create a separate interface for the "Misclassified Images"
misclassified_interface = gr.Interface(show_misclassified_images_wrap,
                    inputs=[gr.Number(value=10, label="Misclassified Inputs",info = "Set the Number of Misclassifed Outputs to be Shown"),
                            gr.Radio(["Yes", "No"], value="No"  , label="Enable GradCAM",info = "Do you want to see GradCAM"),
                            gr.Slider(-2, -1, value=-1, step=1, label="Network Layer",info = "GradCAM Network Layer"),
                            gr.Slider(0, 1, value=0.5, label="Transparency",info = "Set the Opacity of CAM")],
                    outputs=gr.Plot(), description=mislclassified_description)

demo = gr.TabbedInterface([input_interface, misclassified_interface], tab_names=["Top Classes and Prediction", "Misclassified Images"],
                          title=title_with_icon,)
demo.launch()