app.py

import gradio as gr
import random
from PIL import Image, ImageEnhance
import torch
from glob import glob
from torch import nn
from torchvision import transforms

def predict_number(custom_image: Image.Image):

    custom_image = ImageEnhance.Contrast(custom_image.convert("L")).enhance(5.0).point(lambda p: 255 if p > 128 else 0).resize((28, 28))

    transform = transforms.Compose([
    transforms.ToTensor(),
    ])

    model_0.eval()
    with torch.inference_mode():
        pred = torch.softmax(model_0(transform(custom_image).unsqueeze(0).to(device)), dim=1)

    class_names = list("0123456789")
    return {class_names[i]: float(pred[0][i]) for i in range(len(class_names))}

class NumberClassifier(nn.Module):

    def __init__(self, input_shape: int, hidden_units: int, output_shape: int) -> None:
        super().__init__()
        self.conv_block_1 = nn.Sequential(
            nn.Conv2d(in_channels=input_shape, 
                      out_channels=hidden_units, 
                      kernel_size=2, # how big is the square that's going over the image?
                      stride=1, # default
                      ), # options = "valid" (no padding) or "same" (output has same shape as input) or int for specific number 
            nn.ReLU(),
            nn.Conv2d(in_channels=hidden_units, 
                      out_channels=hidden_units,
                      kernel_size=2,
                      stride=1,
                      ),
            nn.ReLU(),
            nn.MaxPool2d(kernel_size=2,
                         stride=2) # default stride value is same as kernel_size
        )
        self.conv_block_2 = nn.Sequential(
            nn.Conv2d(hidden_units, hidden_units, kernel_size=2),
            nn.ReLU(),
            nn.Conv2d(hidden_units, hidden_units, kernel_size=2),
            nn.ReLU(),
            nn.MaxPool2d(2)
        )
        self.classifier = nn.Sequential(
            nn.Flatten(),
            # Where did this in_features shape come from? 
            # It's because each layer of our network compresses and changes the shape of our input data.
            nn.Linear(in_features=hidden_units*5*5,
                      out_features=output_shape)
        )
    
    def forward(self, x: torch.Tensor):
        x = self.conv_block_1(x)
        # print(x.shape)
        x = self.conv_block_2(x)
        # print(x.shape)
        x = self.classifier(x)
        # print(x.shape)
        return x
        # return self.classifier(self.conv_block_2(self.conv_block_1(x))) # <- leverage the benefits of operator fusion

device = "cuda" if torch.cuda.is_available() else "cpu"
torch.manual_seed(42)
model_0 = NumberClassifier(input_shape=1, # number of color channels (3 for RGB) 
                  hidden_units=10, 
                  output_shape=10).to(device)

model_0.load_state_dict(torch.load("models/pytorch_num_classifier_final_model_with_EMNIST.pth", map_location=torch.device('cpu')))


title = "Number Classifier Minimal"
description = "An Image feature extractor computer vision model to classify images of handwritten digits."
article = "Created at [09. PyTorch Model Deployment](https://www.learnpytorch.io/09_pytorch_model_deployment/)."
example_list = [[str(filepath)] for filepath in random.sample(glob("examples/*"), k=25)]
example_list


demo = gr.Interface(fn=predict_number, 
                    inputs=gr.Image(type="pil"),
                    outputs=[gr.Label(num_top_classes=10, label="Predictions")],
                    examples=example_list, 
                    title=title,
                    description=description,
                    article=article)


demo.launch()