app.py

# -*- coding: utf-8 -*-
"""ERAV2-S13-Himank-Gradio.ipynb

Automatically generated by Colab.

Original file is located at
    https://colab.research.google.com/drive/1HJ6wO2_czxZrJwnyUkJ_XaS5HYUvooMS
"""

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
from model import ResNet18

model = ResNet18()
model.load_state_dict(torch.load("model.pth", map_location=torch.device('cpu')), strict=False)

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')

def resize_image_pil(image, new_width, new_height):

    img = Image.fromarray(np.array(image))
    width, height = img.size

    width_scale = new_width / width
    height_scale = new_height / height
    scale = min(width_scale, height_scale)
    resized = img.resize((int(width*scale), int(height*scale)), Image.NEAREST)
    resized = resized.crop((0, 0, new_width, new_height))

    return resized

def inference(input_img,enable_grad_cam,transparency=0.5,target_layer_number=-1,num_top_classes=2):
    input_img = resize_image_pil(input_img, 32, 32)

    input_img = np.array(input_img)
    org_img = input_img
    input_img = input_img.reshape((32, 32, 3))
    transform = transforms.ToTensor()
    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)}
    _, prediction = torch.max(outputs, 1)
    target_layers = [model.layer2[target_layer_number]]
    cam = GradCAM(model=model, target_layers=target_layers)
    grayscale_cam = cam(input_tensor=input_img, targets=None)
    grayscale_cam = grayscale_cam[0, :]
    img = input_img.squeeze(0)
    img = inv_normalize(img)
    if enable_grad_cam:
      visualization = show_cam_on_image(org_img/255, grayscale_cam, use_rgb=True, image_weight=transparency)
    else:
      visualization = None

    confidences = sorted(confidences.items(), key=lambda x: x[1], reverse=True)
    return classes[prediction[0].item()], visualization, dict(confidences[:num_top_classes])

title = "CIFAR10 trained on ResNet18 Model with GradCAM"
description = "A simple Gradio interface to infer on ResNet model, and get GradCAM results"
examples = [
    ["cat.jpg", True, 0.5, -1, 2], ["dog.jpg", True, 0.5, -1, 3], ["bird.jpg", True, 0.5, -1, 4], ["car.jpg", False, 0.5, -1, 5], ["deer.jpg", True, 0.5, -1, 6],
    ["frog.jpg", False, 0.5, -1, 7], ["horse.jpg", False, 0.45, -1, 8], ["plane.jpg", True, 0.30, -2, 9], ["ship.jpg", False, 0.25, -2, 10], ["truck.jpg", True ,0.75, -2, 1]
]
demo = gr.Interface(
    inference,
    inputs = [
        gr.Image(width=256, height=256, label="Input Image"),
        gr.Checkbox(value=False, label="Enable grad-cam image"),
        gr.Slider(0, 1, value = 0.5, label="Overall Opacity of Image"),
        gr.Slider(-2, -1, value = -2, step=1, label="Select Layer"),
        gr.Number(value=2, label="Number of Top Classes to Show", minimum=1, maximum=10),
    ],
    outputs = [
        gr.Textbox(label="Predicted Category"),
        gr.Image(width=256, height=256, label="Output"),
        gr.Label()
    ],
    title = title,
    description = description,
    examples = examples,
)
demo.launch()