datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
MindBigData2022_MNIST_INMindBigData2022_MNIST_MUMindBigData2022_MNIST_EPMNIST-M-flatten
MNIST-M (tabular)
A flattened version of Mike0307/MNIST-M
for tabular learning (XGBoost, AutoGluon, etc.).
Flatten procedure
Each row's image (PNG bytes) is decoded with PIL and converted to RGB.
The 32x32x3 pixels are flattened in row-major (H, W, C) order into 3072 features.
Columns: px_0 ... px_3071 (uint8, 0-255) plus label (int).
The original train/test split is preserved: train.csv, test.csv.
Restoring an image
import pandas as pd, numpy as… See the full description on the dataset page: https://huggingface.co/datasets/Yeongkwon/MNIST-M-flatten.MindBigData2022_MNIST_MWFashion-MNIST-CSVThis dataset is a direct copy of Fashion-MNIST, originally published by Zalando Research on Kaggle https://www.kaggle.com/datasets/zalando-research/fashionmnist.
Fashion-MNIST is a dataset of Zalando's article images—consisting of a training set of 60,000 examples and a test set of 10,000 examples. Each example is a 28x28 grayscale image, associated with a label from 10 classes. Zalando intends Fashion-MNIST to serve as a direct drop-in replacement for the original MNIST dataset for… See the full description on the dataset page: https://huggingface.co/datasets/vincent-espitalier/Fashion-MNIST-CSV.K-MNIST-CSV
Kuzushiji-MNIST
This dataset is a direct CSV conversion of Kuzushiji-MNIST, originally sourced from the GitHub repository https://github.com/rois-codh/kmnist.
Kuzushiji-MNIST is a drop-in replacement for the MNIST dataset (28x28 grayscale, 70,000 images).
fashion-mnistkyrgyz_MNIST
EN: Kyrgyz language hand-written letters (kyrgyz MNIST)
A repository of images (in CSV format) of hand-written Kyrgyz alphabet letters for machine learning applications. Original images have been transformed to 50x50 images and after to csv format.
The repository currently consists of 80213 (50x50 pixel) images representing all 36 letters of the Kyrgyz alphabet These images have been hand-written.
Kaggle competition:
For those who want to compete in Kaggle, HERE… See the full description on the dataset page: https://huggingface.co/datasets/the-cramer-project/kyrgyz_MNIST.mnist_nlp_100mnist_cnnMNISTKyrgyzTest400The data is based on Kyrgyz MNIST.It is based on the Test Set.
Reproduce by:
numSamplesPerCls = 400
seedNum = 512
dfData = pd.read_csv(r'test.csv')
dfT = dfData.groupby('label', group_keys = False).sample(n = numSamplesPerCls, replace = False, random_state = seedNum)
dfT = dfT.reset_index(drop = False)
dfT = dfT.rename(columns = {'index': 'img_index'})
dfT.to_csv(r'MNISTKyrgyzTest400.csv', index = False)
mnist_3tilesMNISTThe dataset contains various MNIST like datasets in teh form of a csv files.
MNIST
Based on the MNIST Dataset in OpenML: OpenML mnist_784.
The way to reproduce:
from sklearn.datasets import fetch_openml
dfX, dsY = fetch_openml('mnist_784', version = 1, return_X_y = True, as_frame = True)
dfX.columns = [str(ii) for ii in range(dfX.shape[1])]
dfX['Label'] = dsY
dfX.to_csv('MNIST.csv')
Fashion MNIST
Based on Zalando Research - FashionMNIST.
Packaged into a CSV in a Row… See the full description on the dataset page: https://huggingface.co/datasets/Royi/MNIST.mnist_4tilesMnist_normalizedMnist normalized data between 0 and 1 instead of 0 and 255
include 1 - hot as label - 10 labels
every sample - 28x28 image pixel number (784)
10 labels
= 794
Mnist_testmnist dataset, values ranges between 0 and 255.
every sample (line) has 794 numbers:
28x28 = 784 - number of pixels in each image (grayscale)
10 - number of labels (in 1 hot)
mnist_tilesmnist