app.py

import gradio as gr
import numpy as np
from PIL import Image
import torch
from torchvision.transforms import ToTensor
from torchvision import transforms
from model import pixact

# Load the ONNX model
model_path = "./model_best.pth"  # Replace with your model path

transforms = transforms.Compose([
    #transforms.Resize(size=(50, 50), antialias=True),
    transforms.ToTensor()
])

net = torch.load(model_path, map_location=torch.device('cpu'))
net = net['arch']
net.eval()
net.cpu()

# Define the superresolution function
def superresolve(image):
  # Preprocess the image
  image = transforms(image)[None,...]



  # Run inference
  output = pixact(net(image))

  # Postprocess the output
  output = output.permute(0,2,3,1)[0].data.numpy()
  output *= 255.0
  output = output.clip(0, 255)
  output = Image.fromarray(np.uint8(output))


  return output

# Define the Gradio interface
interface = gr.Interface(
  fn=superresolve,
  inputs=gr.Image(type="pil"),
  outputs=gr.Image(type="pil"),
  title="Super Resolution",
  description="Upload an image to upscale its resolution.",
  allow_flagging=False,
)

# Launch the Gradio app
interface.launch()

# impath = './image.jpg'
# img = Image.open(impath).convert('RGB')
# superresolve(img)