yiqun/referit
收藏Hugging Face2024-05-09 更新2024-06-12 收录
下载链接：
https://hf-mirror.com/datasets/yiqun/referit
下载链接
链接失效反馈官方服务：
资源简介：
---
license: apache-2.0
---
This repo provides a script to load refcoco, refcocog, and refcoco+ with local files. You should download the data first.
```bash
[DATA_ROOT]
|_ images
|_ train2014
|_ ... (images)
|_ refcoco
|_ instances.json
|_ refs(google).p
|_ refs(unc).p
|_ refcocog
|_ instances.json
|_ refs(google).p
|_ refs(umd).p
|_ refcoco+
|_ instances.json
|_ refs(unc).p
```
Usage:
```python
REPO_NAME = "yiqun/referit"
dataset = "refcoco"
DATA_ROOT = ...
split = "train" # choices: train, val, test, testA, testB
split_by = "unc"
datasets.load_dataset(REPO_NAME, dataset, data_dir=DATA_ROOT, split=split, split_by=split_by)
```
提供机构：
yiqun
原始信息汇总
数据集概述
数据集结构
- images
- train2014
- ... (包含图像文件)
- train2014
- refcoco
- instances.json
- refs(google).p
- refs(unc).p
- refcocog
- instances.json
- refs(google).p
- refs(umd).p
- refcoco+
- instances.json
- refs(unc).p
使用方法
- REPO_NAME: "yiqun/referit"
- dataset: 选择 "refcoco", "refcocog", 或 "refcoco+"
- DATA_ROOT: 数据集根目录
- split: 选择 "train", "val", "test", "testA", 或 "testB"
- split_by: 选择 "unc"
调用方法: python datasets.load_dataset(REPO_NAME, dataset, data_dir=DATA_ROOT, split=split, split_by=split_by)
许可证
- license: apache-2.0
