Hugging Face's logo

Datasets:
nvvarberg
/
chess_pieces

Dataset card Files Community
Dataset Viewer
Viewer
The dataset viewer is not available for this dataset.
The dataset viewer doesn't support this dataset because it runs arbitrary python code. Please open a discussion in the discussion tab if you think this is an error and tag @lhoestq and @severo.
Error code:   DatasetWithScriptNotSupportedError

Need help to make the dataset viewer work? Open a discussion for direct support.

Dataset Description

The Chess Pieces dataset contains images of each chess piece from four separate chess sets. Each piece is photographed in all 64 positions on the board. The camera's perspective is always directly above the center of the board. Images from the first three sets are reserved for training; images from the fourth set are set apart as a test set. Each image is 181 by 181 pixels.

The original dataset is hosted on Zenodo by Pablo Tomas Campos at this link. Note: I am in no way affiliated with Pablo Tomas Campos and had no part in the original data curation. My contribution is simply to restructure and re-post the data on Hugging Face to for accessibility.

Dataset Structure

In total, there are 3,081 images of chess pieces in the dataset. The 767 images from the fourth chess set are set apart as a test set. The other 2,304 images from the first three chess sets make up the training set. Each entry in the dataset contains an image and associated metadata. The full list of data fields is given below.

  • image (Image): PIL image file
  • chess_piece (Categorical): Categorical variable for the name of the chess piece, one of bishop, king, knight, pawn, queen, or rook.
  • location (Categorical): The location on the board the piece is in while photographed. Standard chess locations are used where columns go from A-H and rows go from 1-8.
  • color (Categorical): The color of the chess piece, either black or white.
  • chess_set (Categorical): Which of the four chess sets in the dataset this piece is from.

Processing Steps from Original Dataset

In the original dataset, the metadata for each image is encoded in the file structure and file names. In this dataset, the metadata is linked to each image directly.

Example Use Case

Visit this colab notebook for a tutorial on how to use this dataset for modeling. In the tutorial, I construct an image classification model for the chess pieces.

License

The dataset is published under a CC-BY 4.0 license: https://creativecommons.org/licenses/by/4.0/legalcode.

Citation

Citation for the original dataset hosted on Zenodo:

Pablo Tomas Campos. (2022). Chess piece dataset for image classification (v1.0.0) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.6656212

Downloads last month
0
Edit dataset card