Skip to content

Latest commit

 

History

History
186 lines (154 loc) · 5.92 KB

README-data.md

File metadata and controls

186 lines (154 loc) · 5.92 KB

Data & Setup

This README provides guidelines on how to prepare and structure the datasets. In our paper, we provides 7 CCD benchmarks datasets, i.e., CIFAR100, ImageNet-100, TinyImageNet, and Caltech-101 for generic datasets and Aircraft, Stanford Cars, and CUB for fine-grained datasets.

Dataset

CIFAR100

  1. install cifar2png package using pip. source: link. 
    $ pip install cifar2png
  2. Specifically, you can use below code to download the dataset: 
    $ cifar2png cifar100 data/cifar-100-images --name-with-batch-index
  3. The structure should be: 
    cifar-100-images/
├── test/
└── train/

ImageNet-100

  1. Download the ImageNet dataset from ILSVRC2012.
  2. The structure should be: 
    ILSVRC12/
├── train/
└── val/

TinyImageNet

  1. Download the TinyImageNet dataset from this link.
  2. Unzip the file and the structure should be: 
    tiny-imagenet-200/
├── test/
├── train/
├── val/
├── wnids.txt
└── words.txt

Caltech-101

  1. Download the Caltech-101 dataset from this link.
  2. Unzip the file and the structure should be: 
    caltech-101/
├── accordion/
├── airplanes/
├── ...
├── wrench/
└── yin_yang/

Fine-grained datasets, e.g., Aircraft, Stanford Cars, and CUB

Please follow the instruction here to download these datasets.

*For Stanford Cars dataset, the original download link has been broken, please refer to pytorch/vision#7545 for solution.

Setup

Create data directory and attached symbolic links between the dataset paths and data dir. This can be done by:

*See util/data_util.py for more details / if you want to change the dataset path(s).

# create data directory
$ mkdir data

# attach symbolic links for each dataset path:
# CIFAR100
$ ln -s /dataset/path/cifar-100-images /repository/path/promptccd/data/cifar-100-images

# ImageNet-100
$ ln -s /dataset/path/ILSVRC12 /repository/path/promptccd/data/imagenet

# TinyImageNet
$ ln -s /dataset/path/tiny-imagenet-200 /repository/path/promptccd/data/tiny-imagenet-200

# Caltech-101
$ ln -s /dataset/path/caltech-101 /repository/path/promptccd/data/caltech-101

# Aircraft
$ ln -s /dataset/path/aircraft/fgcv-aircraft-2013b /repository/path/promptccd/data/fgcv-aircraft-2013b

# Stanford Cars
$ ln -s /dataset/path/stanford_car /repository/path/promptccd/data/stanford_car

# CUB
$ ln -s /dataset/path/CUB/CUB_200_2011 /repository/path/promptccd/data/CUB/CUB_200_2011

Citations

The citations for the original datasets (following the order above) are:

%---------------------------------------------------------
% CIFAR100
%---------------------------------------------------------
@article{krizhevsky2009learning,
    title     = {Learning multiple layers of features from tiny images},
    author    = {Krizhevsky, A. and Hinton, G.},
    journal   = {Master's thesis, Department of Computer Science, University of Toronto},
    year      = {2009},
}

%---------------------------------------------------------
% ImageNet-100
%---------------------------------------------------------
@article{russakovsky2015imagenet,
    title     = {Imagenet large scale visual recognition challenge},
    author    = {Russakovsky, Olga and Deng, Jia and Su, Hao and Krause, Jonathan and Satheesh, Sanjeev and Ma, Sean and Huang, Zhiheng and Karpathy, Andrej and Khosla, Aditya and Bernstein, Michael and others},
    journal   = {International Journal of Computer Vision (IJCV)},
    year      = {2015},
}

%---------------------------------------------------------
% TinyImageNet
%---------------------------------------------------------
@article{le2015tiny,
    title     = {Tiny imagenet visual recognition challenge},
    author    = {Le, Ya and Yang, Xuan},
    journal   = {CS 231N},
    year      = {2015},
}

%---------------------------------------------------------
% Caltech-101
%---------------------------------------------------------
@article{fei2006one,
    title     = {One-shot learning of object categories},
    author    = {Fei-Fei, Li and Fergus, Robert and Perona, Pietro},
    journal   = {Transactions on Pattern Analysis and Machine Intelligence (TPAMI)},
    year      = {2006},
    publisher = {IEEE},
}

%---------------------------------------------------------
% Aircraft
%---------------------------------------------------------
@article{maji13fine-grained,
    title    = {Fine-Grained Visual Classification of Aircraft},
    author   = {S. Maji and J. Kannala and E. Rahtu
                        and M. Blaschko and A. Vedaldi},
    journal  = {arXiv preprint arXiv:1306.5151},
    year     = {2013},
}

%---------------------------------------------------------
% Stanford Cars
%---------------------------------------------------------
@inproceedings{krause20133d,
    title     ={3d object representations for fine-grained categorization},
    author    ={Krause, Jonathan and Stark, Michael and Deng, Jia and Fei-Fei, Li},
    booktitle ={4th International IEEE Workshop on  3D Representation and Recognition (3dRR-13)},
    year      ={2013},
}

%---------------------------------------------------------
% CUB
%---------------------------------------------------------
@techreport{WahCUB_200_2011,
    Title       = {{The Caltech-UCSD Birds-200-2011 Dataset}},
    Author      = {Catherine Wah and Steve Branson and Peter Welinder and Pietro Perona and Serge Belongie},
    Year        = {2011},
    Institution = {California Institute of Technology},
}