mirrors/mmengine
1
0
Fork 0
mirror of https://github.com/open-mmlab/mmengine.git synced 2025-06-03 21:54:44 +08:00
Code Issues Projects Releases Wiki Activity

[Doc]: Update the tutorials of basedataset (#445)

Browse Source 
* modify the tutorials of basedataset

* tmp commit
This commit is contained in:
Tao Gong 2022-08-23 11:54:38 +08:00 committed by GitHub
parent ca282aee9e
commit 7fc0a51aae
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 43 additions and 22 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

65
docs/zh_cn/tutorials/basedataset.md
View File

@ -138,10 +138,22 @@ class ToyDataset(BaseDataset):
在定义了数据集类后，就可以通过如下配置实例化 `ToyDataset`： 在定义了数据集类后，就可以通过如下配置实例化 `ToyDataset`：
```python ```python
class LoadImage:
def __call__(self, results):
results['img'] = cv2.imread(results['img_path'])
return results
class ParseImage:
def __call__(self, results):
results['img_shape'] = results['img'].shape
return results
pipeline = [ pipeline = [
dict(type='xxx', ...), LoadImage(),
dict(type='yyy', ...), ParseImage(),
...
] ]
toy_dataset = ToyDataset( toy_dataset = ToyDataset(
@ -154,7 +166,7 @@ toy_dataset = ToyDataset(
同时可以使用数据集类提供的对外接口访问具体的样本信息： 同时可以使用数据集类提供的对外接口访问具体的样本信息：
```python ```python
toy_dataset.meta toy_dataset.metainfo
# dict(classes=('cat', 'dog')) # dict(classes=('cat', 'dog'))
toy_dataset.get_data_info(0) toy_dataset.get_data_info(0)
@ -168,7 +180,13 @@ len(toy_dataset)
# 2 # 2
toy_dataset[0] toy_dataset[0]
# dict(img=xxx, label=0) # {
# 'img_path': "data/train/xxx/xxx_0.jpg",
# 'img_label': 0,
# 'img': a ndarray with shape (H, W, 3), which denotes the value of the image,
# 'img_shape': (H, W, 3) ,
# ...
# }
# `get_subset` 接口不对原数据集类做修改，即完全复制一份新的 # `get_subset` 接口不对原数据集类做修改，即完全复制一份新的
sub_toy_dataset = toy_dataset.get_subset(1) sub_toy_dataset = toy_dataset.get_subset(1)
@ -231,9 +249,8 @@ class ToyVideoDataset(BaseDataset):
```python ```python
pipeline = [ pipeline = [
dict(type='xxx', ...), LoadImage(),
dict(type='yyy', ...), ParseImage(),
...
] ]
toy_dataset = ToyDataset( toy_dataset = ToyDataset(
@ -254,8 +271,16 @@ toy_dataset = ToyDataset(
toy_dataset.full_init() toy_dataset.full_init()
# 初始化完毕，现在可以访问具体数据 # 初始化完毕，现在可以访问具体数据
len(toy_dataset) # 2 len(toy_dataset)
toy_dataset[0] # dict(img=xxx, label=0) # 2
toy_dataset[0]
# {
# 'img_path': "data/train/xxx/xxx_0.jpg",
# 'img_label': 0,
# 'img': a ndarray with shape (H, W, 3), which denotes the value the image,
# 'img_shape': (H, W, 3) ,
# ...
# }
``` ```
**注意：** **注意：**
@ -272,9 +297,8 @@ toy_dataset[0] # dict(img=xxx, label=0)
```python ```python
pipeline = [ pipeline = [
dict(type='xxx', ...), LoadImage(),
dict(type='yyy', ...), ParseImage(),
...
] ]
toy_dataset = ToyDataset( toy_dataset = ToyDataset(
@ -300,9 +324,8 @@ MMEngine 提供了 `ConcatDataset` 包装来拼接多个数据集，使用方法
from mmengine.dataset import ConcatDataset from mmengine.dataset import ConcatDataset
pipeline = [ pipeline = [
dict(type='xxx', ...), LoadImage(),
dict(type='yyy', ...), ParseImage(),
...
] ]
toy_dataset_1 = ToyDataset( toy_dataset_1 = ToyDataset(
@ -331,9 +354,8 @@ MMEngine 提供了 `RepeatDataset` 包装来重复采样某个数据集若干次
from mmengine.dataset import RepeatDataset from mmengine.dataset import RepeatDataset
pipeline = [ pipeline = [
dict(type='xxx', ...), LoadImage(),
dict(type='yyy', ...), ParseImage(),
...
] ]
toy_dataset = ToyDataset( toy_dataset = ToyDataset(
@ -375,9 +397,8 @@ class ToyDataset(BaseDataset):
return [int(data_info['img_label'])] return [int(data_info['img_label'])]
pipeline = [ pipeline = [
dict(type='xxx', ...), LoadImage(),
dict(type='yyy', ...), ParseImage(),
...
] ]
toy_dataset = ToyDataset( toy_dataset = ToyDataset(