app.py

from fastapi import FastAPI
import torch
import torch.nn as nn
import torch
from torchvision import transforms

from typing import Any, Type

import torch

class TorchTensor(torch.Tensor):
 pass

class Prediction():
 prediction: TorchTensor

app = FastAPI()

# Load the PyTorch model
model = torch.load("best_model-epoch=01-val_loss=3.00.ckpt")

# Define a function to preprocess the input


def preprocess_input(input):
 """Preprocess the input image for the PyTorch image classification model.
 Args:
  input: A PIL Image object.
 Returns:
  A PyTorch tensor representing the preprocessed image.
 """

 # Resize the image to the expected size.
 input = input.resize((224, 224))

 # Convert the image to a PyTorch tensor.
 input = torch.from_numpy(np.array(input)).float()

 # Normalize the image.
 input = transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])(input)

 # Return the preprocessed image.
 return input

@app.post("/predict")
async def predict_endpoint(input: Any):
 """Predict the output of the PyTorch image classification model.
 Args:
  input: A file containing the input image.
 Returns:
  A JSON object containing the prediction.
 """

 # Load the image.
 image = Image.open(BytesIO(input))

 # Preprocess the image.
 image = preprocess_input(image)

 # Make a prediction.
 prediction = model(image.unsqueeze(0))

 # Get the top predicted class.
 predicted_class = prediction.argmax(1)

 # Return the prediction.
 return {"prediction": predicted_class.item()}


if _name_ == "_main_":
 import uvicorn
 uvicorn.run(app, host="0.0.0.0", port=8000)