AlexNimoy / nurec

Test System (Neural university recognition check)

Test System (Neural university recognition check)

Colab Examples

Colab Examples.

Usage

!pip install git+https://github.com/AlexNimoy/nurec.git

from nurec import Nurec

recognized_data = [{<JSON_PARAMS>}, {...}]

result = Nurec(recognized_data)

Methods

accuracy

result.accuracy()

returns accuracy like 100.00 or worse

details

result.details()

[key] file
distance: Number
expected:
String
received:
String

total_distance

returns total Levenshtein distance

result.total_distance()

Number

JSON example

{
  "File Name": "",
  "01 Receipt Number": "",
  "02 Receipt Date": "",
  "04 Counterparty": [
    {
      "04-01 Name": "",
      "04-02 BIN": ""
    }
  ],
  "05 Contract": "",
  "06 Our Company BIN": "",
  "07 Counterparty Bank": [
    {
      "07-01 Bank BIC": "",
      "07-02 Current Account IIC": ""
    }
  ],
  "09 Counterparty Payment Purpose Code": "",
  "10 Items Table": [
    {
      "10-01 Item": "",
      "10-02 Unit": "",
      "10-03 Quantity": "",
      "10-04 Price": "",
      "10-05 Amount": ""
    }
  ]
}

Params Table

Params Table (Google Disc)

№	Описание	Название поля
1	Номер Счета	01 Receipt Number
2	Дата Счета	02 Receipt Date
4	Контрагент	04 Counterparty
4.1	Наименование Контрагента	04-01 Name
4.2	БИН Контрагента	04-02 BIN
5	Договор	05 Contract
6	БИН Нашей организации	06 Our Company BIN
7	Банк Контрагента	07 Counterparty Bank
7.1	БИК банка Контрагента	07-01 Bank BIC
7.2	Расчетный счет (ИИК) Контрагента	07-02 Current Account IIC
9	Код назначения платежа Контрагента	09 Counterparty Payment Purpose Code
10	Информация из табличной части:	10 Items Table
10.1	Наименование товара	10-01 Item
10.2	Единица измерения	10-02 Unit
10.3	Количество	10-03 Quantity
10.4	Цена	10-04 Price
10.5	Сумма	10-05 Amount

Скачивание датасета

from nurec import GDriveDataFetcher

# Download dataset
fetcher = GDriveDataFetcher('URL_TO_GOOGLE_DRIVE')
fetcher.download_and_extract()

# Get dataset list
file_list = fetcher.get_file_list()
print(file_list)

Как отправить pull request (Contribute)

Please follow CONTRIBUTING.md.