[Feature] Support some downstream classification datasets.

[zzc98]

Motivation

Add some classification datasets. These datasets are listed below.

dataset	paper	classes	size(train/test)
Oxford 102 Flowers	Automated flower classification over a large number of classes	102	2,040/6,149
Caltech-101	Learning Generative Visual Models from Few Training Examples: An Incremental Bayesian Approach Tested on 101 Object Categories	102	3,060/6,084
Oxford-IIIT Pets	0/1 Deep Neural Networks via Block Coordinate Descent	37	3,680/3,369
Describable Textures (DTD)	Describing Textures in the Wild	47	3,760/1,880
FGVC Aircraft	Fine-Grained Visual Classification of Aircraft	100	6,667/3,333
Stanford Cars	3D Object Representations for Fine-Grained Categorization	196	8,144/8,041
SUN397	SUN Database: Large-scale Scene Recognition from Abbey to Zoo	397	19,850/19,850
Food-101	Food-101 – Mining Discriminative Components with Random Forests	101	75,750/25,250

Examples

Oxford 102 Flowers

>>> from mmpretrain.datasets import Flowers102
>>> train_cfg = dict(data_root='data/Flowers102', split='trainval')
>>> train = Flowers102(**train_cfg)
>>> train
Dataset Flowers102
    Number of samples:  2040
    Root of dataset:    data/Flowers102
>>> test_cfg = dict(data_root='data/Flowers102', split='test')
>>> test = Flowers102(**test_cfg)
>>> test
Dataset Flowers102
    Number of samples:  6149
    Root of dataset:    data/Flowers102

Caltech-101

>>> from mmpretrain.datasets import Caltech101
>>> train_cfg = dict(data_root='data/Caltech', split='train')
>>> train = Caltech101(**train_cfg)
>>> train
Dataset Caltech101
    Number of samples:  3060
    Number of categories:       102
    Root of dataset:    data/Caltech
>>> test_cfg = dict(data_root='data/Caltech', split='test')
>>> test = Caltech101(**test_cfg)
>>> test
Dataset Caltech101
    Number of samples:  6728
    Number of categories:       102
    Root of dataset:    data/Caltech

Oxford-IIIT Pets

>>> from mmpretrain.datasets import OxfordIIITPet
>>> train_cfg = dict(data_root='data/Oxford-IIIT_Pets', split='trainval')
>>> train = OxfordIIITPet(**train_cfg)
>>> train
Dataset OxfordIIITPet
    Number of samples:  3680
    Number of categories:       37
    Root of dataset:    data/Oxford-IIIT_Pets
>>> test_cfg = dict(data_root='data/Oxford-IIIT_Pets', split='test')
>>> test = OxfordIIITPet(**test_cfg)
>>> test
Dataset OxfordIIITPet
    Number of samples:  3669
    Number of categories:       37
    Root of dataset:    data/Oxford-IIIT_Pets

Describable Textures (DTD)

>>> from mmpretrain.datasets import DTD
>>> train_cfg = dict(data_root='data/dtd', split='trainval')
>>> train = DTD(**train_cfg)
>>> train
Dataset DTD
    Number of samples:  3760
    Number of categories:       47
    Root of dataset:    data/dtd
>>> test_cfg = dict(data_root='data/dtd', split='test')
>>> test = DTD(**test_cfg)
>>> test
Dataset DTD
    Number of samples:  1880
    Number of categories:       47
    Root of dataset:    data/dtd

FGVC Aircraft

>>> from mmpretrain.datasets import FGVCAircraft
>>> train_cfg = dict(data_root='data/fgvc-aircraft-2013b/data', split='trainval')
>>> train = FGVCAircraft(**train_cfg)
>>> train
Dataset FGVCAircraft
    Number of samples:  6667
    Number of categories:       100
    Root of dataset:    data/fgvc-aircraft-2013b/data
>>> test_cfg = dict(data_root='data/fgvc-aircraft-2013b/data', split='test')
>>> test = FGVCAircraft(**test_cfg)
>>> test
Dataset FGVCAircraft
    Number of samples:  3333
    Number of categories:       100
    Root of dataset:    data/fgvc-aircraft-2013b/data

Stanford Cars

>>> from mmpretrain.datasets import StanfordCars
>>> train_cfg = dict(data_root='data/Stanford_Cars', split='train')
>>> train = StanfordCars(**train_cfg)
>>> train
Dataset StanfordCars
    Number of samples:  8144
    Number of categories:       196
    Root of dataset:    data/Stanford_Cars
>>> test_cfg = dict(data_root='data/Stanford_Cars', split='test')
>>> test = StanfordCars(**test_cfg)
>>> test
Dataset StanfordCars
    Number of samples:  8041
    Number of categories:       196
    Root of dataset:    data/Stanford_Cars

SUN397

>>> from mmpretrain.datasets import SUN397
>>> train_cfg = dict(data_root='data/SUN397', split='train')
>>> train = SUN397(**train_cfg)
>>> train
Dataset SUN397
    Number of samples:  19824
    Number of categories:       397
    Root of dataset:    data/SUN397
>>> test_cfg = dict(data_root='data/SUN397', split='test')
>>> test = SUN397(**test_cfg)
>>> test
Dataset SUN397
    Number of samples:  19829
    Number of categories:       397
    Root of dataset:    data/SUN397

Food-101

>>> from mmpretrain.datasets import Food101
>>> train_cfg = dict(data_root='data/food-101', split='train')
>>> train = Food101(**train_cfg)
>>> train
Dataset Food101
    Number of samples:  75750
    Number of categories:       101
    Root of dataset:    data/food-101
>>> test_cfg = dict(data_root='data/food-101', split='test')
>>> test = Food101(**test_cfg)
>>> test
Dataset Food101
    Number of samples:  25250
    Number of categories:       101
    Root of dataset:    data/food-101

Checklist

Before PR:

• Pre-commit or other linting tools are used to fix the potential lint issues.
• Bug fixes are fully covered by unit tests, the case that causes the bug should be added in the unit tests.
• The modification is covered by complete unit tests. If not, please add more unit test to ensure the correctness.
• The documentation has been modified accordingly, like docstring or example tutorials.

After PR:

• If the modification has potential influence on downstream or other related projects, this PR should be tested with those projects, like MMDet or MMSeg.
• CLA has been signed and all committers have signed the CLA in this PR.

[codecov]

Codecov Report

Patch coverage has no change and project coverage change: +0.83 🎉

Comparison is base (c9a0cb0) 84.37% compared to head (4c12612) 85.21%.

❗ Current head 4c12612 differs from pull request most recent head ed6c9fe. Consider uploading reports for the commit ed6c9fe to get more accurate results

Additional details and impacted files

@@            Coverage Diff             @@
##              dev    #1467      +/-   ##
==========================================
+ Coverage   84.37%   85.21%   +0.83%     
==========================================
  Files         142      238      +96     
  Lines        9925    17898    +7973     
  Branches     1621     2796    +1175     
==========================================
+ Hits         8374    15251    +6877     
- Misses       1277     2130     +853     
- Partials      274      517     +243     

Flag	Coverage Δ
unittests	85.21% <ø> (+0.83%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

see 380 files with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[wangbo-zhao]

The number of categories in FGVC Aircraft should be 100

[zzc98]

The number of categories in FGVC Aircraft should be 100

The number of categories has been modified. Thanks.

[Ezra-Yu]

please rm the useless files.

[Ezra-Yu]

please recorrect this param doc.

mmpretrain/mmpretrain/datasets/base_dataset.py

Line 52 in e80418a

test_mode (bool): ``test_mode=True`` means in test phase.

[Ezra-Yu]

LGTM.