jxu124/invig

---
language:
- en
- zh
license: apache-2.0
configs:
- config_name: default
data_files:
- split: validation
path: data/validation-*
- split: test
path: data/test-*
- split: train
path: data/train-*
dataset_info:
features:
- name: ref_list
list:
- name: bbox
sequence: float64
- name: category
dtype: string
- name: dialog
sequence:
sequence: string
- name: dialog_cn
sequence:
sequence: string
- name: id
dtype: string
- name: image_info
struct:
- name: file_name
dtype: string
- name: height
dtype: int64
- name: id
dtype: string
- name: width
dtype: int64
- name: image
dtype: image
splits:
- name: validation
num_bytes: 96380848.0
num_examples: 996
- name: test
num_bytes: 193325330.698
num_examples: 1997
- name: train
num_bytes: 1735786813.55
num_examples: 17710
download_size: 865015922
dataset_size: 2025492992.248
---
# Dataset Card for "invig"

[Github](https://github.com/ZhangHanbo/invig-dataset)

```latex
@misc{invigdataset,
title={InViG: Interactive Visual-Language Disambiguation with 21K Human-to-Human Dialogues},
author={Zhang, Hanbo and Mo, Yuchen and Xu, Jie and Si, Qingyi and Kong, Tao},
howpublished = {\url{https://github.com/ZhangHanbo/invig-dataset}},
year={2023}
}
```

原始信息汇总

数据集概述

数据集信息

• 语言: 英语（en）、中文（zh）
• 许可证: Apache-2.0

配置

• 默认配置 (config_name: default)
  • 数据文件路径:
    • 验证集 (split: validation): data/validation-*
    • 测试集 (split: test): data/test-*
    • 训练集 (split: train): data/train-*

数据集特征

• ref_list
  • bbox: 浮点数序列
  • category: 字符串
  • dialog: 字符串序列
  • dialog_cn: 字符串序列
  • id: 字符串
• image_info
  • file_name: 字符串
  • height: 整数
  • id: 字符串
  • width: 整数
• image: 图像类型

数据集分割

• 验证集 (name: validation)
  • 示例数量: 996
  • 数据大小: 96380848.0 字节
• 测试集 (name: test)
  • 示例数量: 1997
  • 数据大小: 193325330.698 字节
• 训练集 (name: train)
  • 示例数量: 17710
  • 数据大小: 1735786813.55 字节

数据集大小

• 下载大小: 865015922 字节
• 数据集总大小: 2025492992.248 字节