jp1924/OCRDataPublic-inst

---
dataset_info:
features:
- name: image
dtype: image
- name: meta_data
struct:
- name: object_recognition
dtype: int32
- name: text_language
dtype: int32
- name: category
dtype: int32
- name: identifier
dtype: string
- name: label_path
dtype: string
- name: name
dtype: string
- name: src_path
dtype: string
- name: type
dtype: string
- name: acquisition_location
dtype: int32
- name: data_captured
dtype: string
- name: dpi
dtype: int32
- name: group
dtype: int32
- name: height
dtype: int32
- name: width
dtype: int32
- name: writing_style
dtype: int32
- name: year
dtype: int32
- name: objects
list:
- name: id
dtype: int32
- name: text
dtype: string
- name: bbox
list: int32
- name: meta
struct:
- name: type
dtype: string
- name: text_type
dtype: string
- name: context
dtype: string
- name: question
sequence: string
- name: answer
sequence: string
splits:
- name: train
num_bytes: 14019183278.0
num_examples: 40000
- name: validation
num_bytes: 1066622883.0
num_examples: 3000
download_size: 10970846438
dataset_size: 15085806161.0
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*
---

原始信息汇总

数据集概述

数据集特征

• image: 图像数据类型。
• meta_data: 结构化数据，包含以下字段：
  • object_recognition: int32
  • text_language: int32
  • category: int32
  • identifier: string
  • label_path: string
  • name: string
  • src_path: string
  • type: string
  • acquisition_location: int32
  • data_captured: string
  • dpi: int32
  • group: int32
  • height: int32
  • width: int32
  • writing_style: int32
  • year: int32
• objects: 列表数据，包含以下字段：
  • id: int32
  • text: string
  • bbox: 列表，包含int32类型数据
  • meta: 结构化数据，包含以下字段：
    • type: string
    • text_type: string
• context: string
• question: 序列，包含string类型数据
• answer: 序列，包含string类型数据

数据集分割

• train: 40000个样本，总大小为14019183278字节。
• validation: 3000个样本，总大小为1066622883字节。

数据集大小

• 下载大小: 10970846438字节
• 数据集总大小: 15085806161字节

数据文件配置

• default 配置:
  • train: 路径为data/train-*
  • validation: 路径为data/validation-*