app.py

import torch
from PIL import Image
from strhub.data.module import SceneTextDataModule
from strhub.models.utils import load_from_checkpoint, parse_model_args
import gradio as gr


# # Load model and image transforms
# parseq = torch.hub.load('baudm/parseq', 'trba', pretrained=True).eval()
from strhub.models.crnn.system import CRNN as ModelClass
parseq = ModelClass.load_from_checkpoint("sneed.ckpt").eval()

img_transform = SceneTextDataModule.get_transform(parseq.hparams.img_size)

def captcha_solver(img):
  img = img.convert('RGB')
  img = img_transform(img).unsqueeze(0)

  logits = parseq(img)
  logits.shape
  
  # # Greedy decoding
  pred = logits.softmax(-1)
  label, confidence = parseq.tokenizer.decode(pred)
  return label[0]

demo = gr.Interface(fn=captcha_solver, inputs=gr.inputs.Image(type="pil"), outputs=gr.outputs.Textbox())
demo.launch()