eurecom-ds/shapes3d-dist

---
dataset_info:
- config_name: shifted_high_distribution
features:
- name: image
dtype: image
- name: label_floor_hue
dtype:
class_label:
names:
'0': '0'
'1': '1'
'2': '2'
'3': '3'
'4': '4'
'5': '5'
'6': '6'
'7': '7'
'8': '8'
'9': '9'
- name: label_object_hue
dtype:
class_label:
names:
'0': '0'
'1': '1'
'2': '2'
'3': '3'
'4': '4'
'5': '5'
'6': '6'
'7': '7'
'8': '8'
'9': '9'
- name: label_orientation
dtype:
class_label:
names:
'0': '0'
'1': '1'
'2': '2'
'3': '3'
'4': '4'
'5': '5'
'6': '6'
'7': '7'
'8': '8'
'9': '9'
'10': '10'
'11': '11'
'12': '12'
'13': '13'
'14': '14'
- name: label_scale
dtype:
class_label:
names:
'0': '0'
'1': '1'
'2': '2'
'3': '3'
'4': '4'
'5': '5'
'6': '6'
'7': '7'
- name: label_shape
dtype:
class_label:
names:
'0': '0'
'1': '1'
'2': '2'
'3': '3'
- name: label_wall_hue
dtype:
class_label:
names:
'0': '0'
'1': '1'
'2': '2'
'3': '3'
'4': '4'
'5': '5'
'6': '6'
'7': '7'
'8': '8'
'9': '9'
- name: value_floor_hue
dtype: float32
- name: value_object_hue
dtype: float32
- name: value_orientation
dtype: float32
- name: value_scale
dtype: float32
- name: value_shape
dtype: float32
- name: value_wall_hue
dtype: float32
splits:
- name: train
num_bytes: 1047792.814375
num_examples: 300
download_size: 1086457
dataset_size: 1047792.814375
- config_name: shifted_low_distribution
features:
- name: image
dtype: image
- name: label_floor_hue
dtype:
class_label:
names:
'0': '0'
'1': '1'
'2': '2'
'3': '3'
'4': '4'
'5': '5'
'6': '6'
'7': '7'
'8': '8'
'9': '9'
- name: label_object_hue
dtype:
class_label:
names:
'0': '0'
'1': '1'
'2': '2'
'3': '3'
'4': '4'
'5': '5'
'6': '6'
'7': '7'
'8': '8'
'9': '9'
- name: label_orientation
dtype:
class_label:
names:
'0': '0'
'1': '1'
'2': '2'
'3': '3'
'4': '4'
'5': '5'
'6': '6'
'7': '7'
'8': '8'
'9': '9'
'10': '10'
'11': '11'
'12': '12'
'13': '13'
'14': '14'
- name: label_scale
dtype:
class_label:
names:
'0': '0'
'1': '1'
'2': '2'
'3': '3'
'4': '4'
'5': '5'
'6': '6'
'7': '7'
- name: label_shape
dtype:
class_label:
names:
'0': '0'
'1': '1'
'2': '2'
'3': '3'
- name: label_wall_hue
dtype:
class_label:
names:
'0': '0'
'1': '1'
'2': '2'
'3': '3'
'4': '4'
'5': '5'
'6': '6'
'7': '7'
'8': '8'
'9': '9'
- name: value_floor_hue
dtype: float32
- name: value_object_hue
dtype: float32
- name: value_orientation
dtype: float32
- name: value_scale
dtype: float32
- name: value_shape
dtype: float32
- name: value_wall_hue
dtype: float32
splits:
- name: train
num_bytes: 1047792.814375
num_examples: 300
download_size: 1058998
dataset_size: 1047792.814375
- config_name: uniform_distribution
features:
- name: image
dtype: image
- name: label_floor_hue
dtype:
class_label:
names:
'0': '0'
'1': '1'
'2': '2'
'3': '3'
'4': '4'
'5': '5'
'6': '6'
'7': '7'
'8': '8'
'9': '9'
- name: label_object_hue
dtype:
class_label:
names:
'0': '0'
'1': '1'
'2': '2'
'3': '3'
'4': '4'
'5': '5'
'6': '6'
'7': '7'
'8': '8'
'9': '9'
- name: label_orientation
dtype:
class_label:
names:
'0': '0'
'1': '1'
'2': '2'
'3': '3'
'4': '4'
'5': '5'
'6': '6'
'7': '7'
'8': '8'
'9': '9'
'10': '10'
'11': '11'
'12': '12'
'13': '13'
'14': '14'
- name: label_scale
dtype:
class_label:
names:
'0': '0'
'1': '1'
'2': '2'
'3': '3'
'4': '4'
'5': '5'
'6': '6'
'7': '7'
- name: label_shape
dtype:
class_label:
names:
'0': '0'
'1': '1'
'2': '2'
'3': '3'
- name: label_wall_hue
dtype:
class_label:
names:
'0': '0'
'1': '1'
'2': '2'
'3': '3'
'4': '4'
'5': '5'
'6': '6'
'7': '7'
'8': '8'
'9': '9'
- name: value_floor_hue
dtype: float32
- name: value_object_hue
dtype: float32
- name: value_orientation
dtype: float32
- name: value_scale
dtype: float32
- name: value_shape
dtype: float32
- name: value_wall_hue
dtype: float32
splits:
- name: train
num_bytes: 1047792.814375
num_examples: 300
download_size: 1073972
dataset_size: 1047792.814375
configs:
- config_name: shifted_high_distribution
data_files:
- split: train
path: shifted_high_distribution/train-*
- config_name: shifted_low_distribution
data_files:
- split: train
path: shifted_low_distribution/train-*
- config_name: uniform_distribution
data_files:
- split: train
path: uniform_distribution/train-*
---

数据集信息如下：

### 配置1：高偏移分布（shifted_high_distribution）
本配置包含以下特征：
1. `image`：数据类型为图像（image）
2. `label_floor_hue`：分类标签（class_label）类型，标签映射为{'0': '0', '1': '1', '2': '2', '3': '3', '4': '4', '5': '5', '6': '6', '7': '7', '8': '8', '9': '9'}，共10个类别
3. `label_object_hue`：分类标签类型，标签映射与`label_floor_hue`一致，共10个类别
4. `label_orientation`：分类标签类型，标签映射为{'0': '0', '1': '1', '2': '2', '3': '3', '4': '4', '5': '5', '6': '6', '7': '7', '8': '8', '9': '9', '10': '10', '11': '11', '12': '12', '13': '13', '14': '14'}，共15个类别
5. `label_scale`：分类标签类型，标签映射为{'0': '0', '1': '1', '2': '2', '3': '3', '4': '4', '5': '5', '6': '6', '7': '7'}，共8个类别
6. `label_shape`：分类标签类型，标签映射为{'0': '0', '1': '1', '2': '2', '3': '3'}，共4个类别
7. `label_wall_hue`：分类标签类型，标签映射与`label_floor_hue`一致，共10个类别
8. `value_floor_hue`：32位浮点数据类型（float32）
9. `value_object_hue`：32位浮点数据类型
10. `value_orientation`：32位浮点数据类型
11. `value_scale`：32位浮点数据类型
12. `value_shape`：32位浮点数据类型
13. `value_wall_hue`：32位浮点数据类型

数据集仅包含训练集（train）划分：字节大小为1047792.814375，样本总数300。本配置下载大小为1086457，数据集总大小为1047792.814375。

### 配置2：低偏移分布（shifted_low_distribution）
本配置特征与高偏移分布配置完全一致。数据集仅包含训练集划分：字节大小为1047792.814375，样本总数300。本配置下载大小为1058998，数据集总大小为1047792.814375。

### 配置3：均匀分布（uniform_distribution）
本配置特征与前述两个配置完全一致。数据集仅包含训练集划分：字节大小为1047792.814375，样本总数300。本配置下载大小为1073972，数据集总大小为1047792.814375。

### 全局数据集配置
各配置对应的训练集数据文件路径如下：
1. 高偏移分布（shifted_high_distribution）：`shifted_high_distribution/train-*`
2. 低偏移分布（shifted_low_distribution）：`shifted_low_distribution/train-*`
3. 均匀分布（uniform_distribution）：`uniform_distribution/train-*`