Data
Meta_Album_AWA_Extended

Meta_Album_AWA_Extended

active ARFF CC BY-NC 4.0 Visibility: public Uploaded 08-11-2022 by Meta Album
0 likes downloaded by 0 people , 0 total downloads 0 issues 0 downvotes
Issue #Downvotes for this reason By


Loading wiki
Help us complete this description Edit
## Meta-Album Animals with Attributes Dataset (Extended) * The original Animals with Attributes 2 (AWA) dataset (https://cvml.ist.ac.at/AwA2/) was designed to benchmark transfer-learning algorithms, in particular attribute base classification and zero-shot learning. It has more than 37 000 images from 50 animals, where each animal corresponds to a class. The images of this dataset were collected from public sources, such as Flickr, in 2016, considering only images licensed for free use and redistribution. Each class can have 100 to 1 645 images with a resolution from 100x100 to 1 893x1 920 px. To preprocess this dataset, we cropped the images from either side to make them square. In case an image has a resolution lower than 128 px, the squared images are done by either duplicating the top and bottom-most 3 rows or the left and right most 3 columns based on the orientation of the original image. Lastly, the square images are resized into 128x128 px using an anti-aliasing filter. ### Dataset Details ![](https://meta-album.github.io/assets/img/samples/AWA.png) Meta Album ID: LR_AM.AWA Meta Album URL: [https://meta-album.github.io/datasets/AWA.html](https://meta-album.github.io/datasets/AWA.html) Domain ID: LR_AM Domain Name: Large Aninamls Dataset ID: AWA Dataset Name: Animals with Attributes Short Description: Mamals dataset for image classification \# Classes: 50 \# Images: 37318 Keywords: mammals, animals, Data Format: images Image size: 128x128 License (original data release): Creative Commons License URL(original data release): https://cvml.ist.ac.at/AwA2/ License (Meta-Album data release): Creative Commons License URL (Meta-Album data release): [https://cvml.ist.ac.at/AwA2/](https://cvml.ist.ac.at/AwA2/) Source: Animals with attributes 2 Source URL: https://cvml.ist.ac.at/AwA2/ Original Author: Christoph H. Lampert, Bernt Schiele, Zeynep Akata Original contact: chl@ist.ac.at Meta Album author: Dustin Carrion Created Date: 01 March 2022 Contact Name: Ihsan Ullah Contact Email: meta-album@chalearn.org Contact URL: [https://meta-album.github.io/](https://meta-album.github.io/) ### Cite this dataset ``` @ARTICLE{8413121, author={Xian, Yongqin and Lampert, Christoph H. and Schiele, Bernt and Akata, Zeynep}, journal={IEEE Transactions on Pattern Analysis and Machine Intelligence}, title={Zero-Shot Learning - A Comprehensive Evaluation of the Good, the Bad and the Ugly}, year={2019}, volume={41}, number={9}, pages={2251-2265}, doi={10.1109/TPAMI.2018.2857768} } ``` ### Cite Meta-Album ``` @inproceedings{meta-album-2022, title={Meta-Album: Multi-domain Meta-Dataset for Few-Shot Image Classification}, author={Ullah, Ihsan and Carrion, Dustin and Escalera, Sergio and Guyon, Isabelle M and Huisman, Mike and Mohr, Felix and van Rijn, Jan N and Sun, Haozhe and Vanschoren, Joaquin and Vu, Phan Anh}, booktitle={Thirty-sixth Conference on Neural Information Processing Systems Datasets and Benchmarks Track}, url = {https://meta-album.github.io/}, year = {2022} } ``` ### More For more information on the Meta-Album dataset, please see the [[NeurIPS 2022 paper]](https://meta-album.github.io/paper/Meta-Album.pdf) For details on the dataset preprocessing, please see the [[supplementary materials]](https://openreview.net/attachment?id=70_Wx-dON3q&name=supplementary_material) Supporting code can be found on our [[GitHub repo]](https://github.com/ihsaan-ullah/meta-album) Meta-Album on Papers with Code [[Meta-Album]](https://paperswithcode.com/dataset/meta-album) ### Other versions of this dataset [[Micro]](https://www.openml.org/d/44275) [[Mini]](https://www.openml.org/d/44305)

3 features

CATEGORY (target)string50 unique values
0 missing
FILE_NAMEstring37318 unique values
0 missing
SUPER_CATEGORYnumeric0 unique values
37318 missing

19 properties

37318
Number of instances (rows) of the dataset.
3
Number of attributes (columns) of the dataset.
50
Number of distinct values of the target attribute (if it is nominal).
37318
Number of missing values in the dataset.
37318
Number of instances with at least one value missing.
1
Number of numeric attributes.
0
Number of nominal attributes.
0
Percentage of binary attributes.
100
Percentage of instances having missing values.
33.33
Percentage of missing values.
1
Average class difference between consecutive instances.
33.33
Percentage of numeric attributes.
0
Number of attributes divided by the number of instances.
0
Percentage of nominal attributes.
4.41
Percentage of instances belonging to the most frequent class.
1644
Number of instances belonging to the most frequent class.
0.27
Percentage of instances belonging to the least frequent class.
100
Number of instances belonging to the least frequent class.
0
Number of binary attributes.

1 tasks

0 runs - estimation_procedure: 10-fold Crossvalidation - evaluation_measure: predictive_accuracy - target_feature: CATEGORY
Define a new task