Spaces:
Sleeping
Sleeping
File size: 5,404 Bytes
70085b1 3d858b9 70085b1 |
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 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 |
import gradio as gr
import requests
import json
from jiwer import cer, wer
import re
pdf_file_path = 'dummy.pdf'
with open("page_transcriptions.json", encoding="utf-8") as f:
data = json.load(f)
def send_request(url):
try:
with open(pdf_file_path, 'rb') as pdf_file:
files = {
'file': (
pdf_file_path,
pdf_file,
'application/pdf'
)
}
response = requests.post(url, files=files)
except Exception as e:
return {"Error message: "f"Error occurred while sending request. Error message: {e}"}
try:
response_json = response.json()
except Exception as e:
return {
"Error message": e,
"Response": response.content
}
if isinstance(response_json, list):
for page in response_json:
if isinstance(page, dict):
if "page_number" not in page.keys() or "MD_text" not in page.keys():
return {
"Error message": "Response is not in desired structure. Desired structure: [{'page_number': 1, 'MD_text': 'Extracted text'}]",
"Response": response_json
}
if isinstance(page["page_number"], int) and isinstance(page["MD_text"], str):
continue
else:
return {
"Error message": "'page_number' should be integer and 'MD_text' should be string.",
"Response": response_json
}
else:
return {
"Error message": "List should include only dictionaries.",
"Response": response_json
}
if len(response_json) != len(data):
return {
"Error message": "The number of pages are not equal between transcription and ground truth.",
"Response": response_json
}
final_metrics = []
total_reference = ""
total_hypothesis = ""
for page in response_json:
for transcription in data:
if page["page_number"] == transcription["page_number"]:
reference = transcription['MD_text'].strip()
hypothesis = page['MD_text'].strip()
reference = reference.lower()
hypothesis = hypothesis.lower()
reference = reference.replace("\n", " ")
hypothesis = hypothesis.replace("\n", " ")
reference = re.sub(r'\s+', ' ', reference)
hypothesis = re.sub(r'\s+', ' ', hypothesis)
total_reference += reference
total_reference += " "
total_hypothesis += hypothesis
total_hypothesis += " "
cer_value = max(1 - cer(reference, hypothesis), 0)
wer_value = max(1 - wer(reference, hypothesis), 0)
final_metrics.append({"page_number": page["page_number"], "Character Success Rate (CSR)": round(cer_value, 4), "Word Success Rate (WSR)": round(wer_value, 4), "MD_text_used_for_metrics": hypothesis, "Ground_Truth_used_for_metrics": reference})
global_cer = max(1 - cer(total_reference.strip(), total_hypothesis.strip()), 0)
global_wer = max(1 - wer(total_reference.strip(), total_hypothesis.strip()), 0)
final_metrics.append({"Global CSR": global_cer, "Global WSR": global_wer, "MD_text_used_for_metrics": total_hypothesis.strip(), "Ground_Truth_used_for_metrics": total_reference.strip()})
return final_metrics
else:
return {
"Error message": "Response should be list of dictionaries.",
"Response": response_json
}
with gr.Blocks() as demo:
gr.Markdown(
"""
# OCR Endpoint Response Validator and Quality Checker
Character Success Rate (CSR) and Word Success Rate (WSR) are metrics that will be provided for each page and total.
They are calculated by simply subtracting CER and WER from 1 respectively.
If CER or WER is > 1, CSR or WSR is considered as 0.
Enter your endpoint below and click **Send** to get the result.
Format:
```http://<host>/<endpoint>```
"""
)
output = gr.JSON(
label="Output"
)
input_box = gr.Textbox(
label="Input",
lines=1,
placeholder="Type your text here..."
)
send_btn = gr.Button("Send")
send_btn.click(
fn=send_request,
inputs=input_box,
outputs=output
)
input_box.submit(
fn=send_request,
inputs=input_box,
outputs=output
)
demo.launch() |