File size: 1,480 Bytes
e5bfa19
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
47d96e7
 
 
 
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
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
---
language:
- en
library_name: transformers
tags:
- donut
- donut-python
---

### Installtion
```bash
pip install torch 
pip install transformers==4.11.3
pip install opencv-python==4.6.0.66
pip install donut-python
```

### Usage 
```python
import sys
import os
import pandas as pd
import numpy as np
import shutil

from tqdm import tqdm
import re

from donut import DonutModel
import torch
from PIL import Image

en_model_path = "question_generator_by_en_on_pic"

task_prompt = "<s_docvqa><s_question>{user_input}</s_question><s_answer>"
en_pretrained_model = DonutModel.from_pretrained(en_model_path)

if torch.cuda.is_available():
    en_pretrained_model.half()
    device = torch.device("cuda")
    en_pretrained_model.to(device)

en_pretrained_model.eval()
print("have load !")

def demo_process_vqa(input_img, question):
    #input_img = Image.fromarray(input_img)
    global en_pretrained_model, task_prompt
    user_prompt = task_prompt.replace("{user_input}", question)
    output = en_pretrained_model.inference(input_img, prompt=user_prompt)["predictions"][0]
    req = {
        "question": output["answer"],
        "answer": output["question"]
    }
    return req


img_path = "en_img.png"
demo_process_vqa(Image.open(img_path), "605-7227", "en")

'''
{
  "question": "What is the Phone #?",
  "answer": "605-7227"
}
'''

```

### Sample Image 
<img src="https://raw.githubusercontent.com/svjack/docvqa-gen/main/imgs/en_img.png" width = "500px" height = "500px"/>