RoScenes-release

## Dataset Card of RoScenes

For detailed information, please refer to https://roscenes.github.io and https://github.com/roscenes/RoScenes

### Download instructions

**NOTE: Storage usage: [train: 370GB / validation: 95 GB / test: 150GB]**

1. Download all the raw `[SPLIT]/[SPLIT]-parts[N].tar` into your local storage (`SPLIT` could be `train/validation/test`), you can follow the official instructions:

:modelscope-code[]{type="sdk"}
:modelscope-code[]{type="git"}

2. If you just want to download partial of files, you could use the `skip-smudge` feature of Git LFS:

```bash
# skip cloning LFS files
GIT_LFS_SKIP_SMUDGE=1 git clone https://www.modelscope.cn/datasets/Apsara_Lab_Multimodal_Intelligence/RoScenes-release.git

cd RoScenes-release

# init Git LFS
git lfs install --skip-smudge

# manually pull files you want, for example, --include="train/*"
git pull --include=[PATTERN]
```

3. Prepare another folder as the `DATA_ROOT`.
4. Then, extract the tars one by one, do not use `cat`. For example, for the training set:
```bash
#!/bin/bash
for file in train/train-parts{0..19}.tar; do
echo "extracting $file...";
tar -xf $file -C [DATA_ROOT];
done
```
5. Check the resulting `DATA_ROOT`, it should be organized as follows:

```yaml
. [DATA_ROOT] # Dataset root folder
├── 📂train # training set
│ ├── 📂s001_split_train_difficulty_mixed_ambience_day # scene 001's data
│ │ ├── 📂database # annotations, grouped by clip
│ │ │ ├── 📂0076fd69_clip_[0000000000000-0000000029529] # a clip's database, please use our devkit to read
│ │ │ └ ...
│ │ └── 📂images # images, grouped by clips
│ │ ├── 📂0076fd69
│ │ └ ...
│ ├── 📂s002_split_train_difficulty_mixed_ambience_day
│ ├── 📂s003_split_train_difficulty_mixed_ambience_day
│ ├── 📂s004_split_train_difficulty_mixed_ambience_day
│ └── 📂night_split_train_difficulty_mixed_ambience_night
│
│
├── 📂validation # validation set
│ ├── 📂s001_split_validation_difficulty_mixed_ambience_day # scene 001's data
│ ├── 📂s002_split_validation_difficulty_mixed_ambience_day
│ ├── 📂s003_split_validation_difficulty_mixed_ambience_day
│ ├── 📂s004_split_validation_difficulty_mixed_ambience_day
│ └── 📂night_split_validation_difficulty_mixed_ambience_night
│
│
└── 📂test # test set
├── 📂NO_GTs005_split_test_difficulty_mixed_ambience_day # scene 005's data
├── 📂NO_GTs006_split_test_difficulty_mixed_ambience_day
├── 📂NO_GTs007_split_test_difficulty_mixed_ambience_day
├── 📂NO_GTs008_split_test_difficulty_mixed_ambience_day
├── 📂NO_GTs009_split_test_difficulty_mixed_ambience_day
├── 📂NO_GTs010_split_test_difficulty_mixed_ambience_day
├── 📂NO_GTs011_split_test_difficulty_mixed_ambience_day
├── 📂NO_GTs012_split_test_difficulty_mixed_ambience_day
├── 📂NO_GTs013_split_test_difficulty_mixed_ambience_day
└── 📂NO_GTs014_split_test_difficulty_mixed_ambience_day

```

### Note

Commercial use of RoScenes is strictly forbidden.

# RoScenes数据集卡片
如需获取详细信息，请访问 https://roscenes.github.io 与 https://github.com/roscenes/RoScenes

## 下载指引
**注意：存储空间占用：[训练集：370GB / 验证集：95GB / 测试集：150GB]**

1. 将所有`[SPLIT]/[SPLIT]-parts[N].tar`原始文件下载至本地存储（`SPLIT`可选取值为`train/validation/test`），您可遵循官方指引操作：
:modelscope-code[]{type="sdk"}
:modelscope-code[]{type="git"}

2. 若仅需下载部分文件，可使用Git大文件存储（Git LFS）的`skip-smudge`功能：
bash
# 跳过克隆LFS文件
GIT_LFS_SKIP_SMUDGE=1 git clone https://www.modelscope.cn/datasets/Apsara_Lab_Multimodal_Intelligence/RoScenes-release.git

cd RoScenes-release

# 初始化Git LFS
git lfs install --skip-smudge

# 手动拉取所需文件，例如 --include="train/*"
git pull --include=[PATTERN]

3. 新建独立文件夹作为`DATA_ROOT`。
4. 随后请逐个解压所有tar文件，请勿使用`cat`命令。以训练集为例：
bash
#!/bin/bash
for file in train/train-parts{0..19}.tar; do
echo "extracting $file...";
tar -xf $file -C [DATA_ROOT];
done

5. 检查生成的`DATA_ROOT`，其目录组织格式应如下所示：
yaml
. [DATA_ROOT] # 数据集根文件夹
├── 📂train # 训练集
│ ├── 📂s001_split_train_difficulty_mixed_ambience_day # 场景001的数据
│ │ ├── 📂database # 按剪辑分组的标注文件
│ │ │ ├── 📂0076fd69_clip_[0000000000000-0000000029529] # 单个剪辑的标注数据，请使用我们的开发工具包读取
│ │ │ └ ...
│ │ └── 📂images # 按剪辑分组的图像文件
│ │ ├── 📂0076fd69
│ │ └ ...
│ ├── 📂s002_split_train_difficulty_mixed_ambience_day
│ ├── 📂s003_split_train_difficulty_mixed_ambience_day
│ ├── 📂s004_split_train_difficulty_mixed_ambience_day
│ └── 📂night_split_train_difficulty_mixed_ambience_night
│
│
├── 📂validation # 验证集
│ ├── 📂s001_split_validation_difficulty_mixed_ambience_day # 场景001的数据
│ ├── 📂s002_split_validation_difficulty_mixed_ambience_day
│ ├── 📂s003_split_validation_difficulty_mixed_ambience_day
│ ├── 📂s004_split_validation_difficulty_mixed_ambience_day
│ └── 📂night_split_validation_difficulty_mixed_ambience_night
│
│
└── 📂test # 测试集
├── 📂NO_GTs005_split_test_difficulty_mixed_ambience_day # 场景005的数据（无标注）
├── 📂NO_GTs006_split_test_difficulty_mixed_ambience_day
├── 📂NO_GTs007_split_test_difficulty_mixed_ambience_day
├── 📂NO_GTs008_split_test_difficulty_mixed_ambience_day
├── 📂NO_GTs009_split_test_difficulty_mixed_ambience_day
├── 📂NO_GTs010_split_test_difficulty_mixed_ambience_day
├── 📂NO_GTs011_split_test_difficulty_mixed_ambience_day
├── 📂NO_GTs012_split_test_difficulty_mixed_ambience_day
├── 📂NO_GTs013_split_test_difficulty_mixed_ambience_day
└── 📂NO_GTs014_split_test_difficulty_mixed_ambience_day

## 注意事项
严禁将RoScenes用于商业用途。