File size: 2,286 Bytes
cc9f92c 77fa895 cc9f92c 1172c31 cc9f92c c6e3997 0cdacaf cc9f92c c6e3997 cc9f92c 1172c31 8094952 cc9f92c 61546ab cc9f92c 8094952 cc9f92c 8094952 cc9f92c |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 |
import torch, re
from PIL import Image
from transformers import DonutProcessor, VisionEncoderDecoderModel
import streamlit as st
from dotenv import load_dotenv
import os
import time
load_dotenv()
# image_path = '/app/Datasplit/test/1099_Div/filled_form_43.jpg'
# image = Image.open(image_path)
# imgae = image.resize((1864, 1440))
device = "cuda" if torch.cuda.is_available() else "cpu"
# device = "cpu"
# Load the processor from the local directory
processor = DonutProcessor.from_pretrained("Henge-navuuu/donut-base-finetuned-forms-v1")
# processor.to(device)
# Load the model from the local directory
model = VisionEncoderDecoderModel.from_pretrained("Henge-navuuu/donut-base-finetuned-forms-v1")
model.to(device)
@st.cache_resource
def inference(image):
pixel_values = processor(image, return_tensors="pt").pixel_values
task_prompt = "<s>"
decoder_input_ids = processor.tokenizer(task_prompt, add_special_tokens=False, return_tensors="pt")["input_ids"]
# device = "cuda" if torch.cuda.is_available() else "cpu"
# model.to(device)
start_time = time.time()
outputs = model.generate(pixel_values.to(device),
decoder_input_ids=decoder_input_ids.to(device),
max_length=model.decoder.config.max_position_embeddings,
early_stopping=True,
pad_token_id=processor.tokenizer.pad_token_id,
eos_token_id=processor.tokenizer.eos_token_id,
use_cache=True,
num_beams=2,
bad_words_ids=[[processor.tokenizer.unk_token_id]],
return_dict_in_generate=True,
output_scores=True,)
end_time = time.time()
sequence = processor.batch_decode(outputs.sequences)[0]
sequence = sequence.replace(processor.tokenizer.eos_token, "").replace(processor.tokenizer.pad_token, "")
sequence = re.sub(r"<.*?>", "", sequence, count=1).strip() # remove first task start token
print(processor.token2json(sequence))
print(f"Donut Inference time {start_time-end_time}")
return processor.token2json(sequence)
# data = inference(image)
# print(data) |