• Description:

A re-labeled version of CIFAR-100 with real human annotation errors. For every pair (image, label) in the original CIFAR-100 train set, it provides an additional label given by a real human annotator.

Then convert 'CIFAR-100_human_ordered.npy' into a CSV file 'CIFAR-100_human_annotations.csv'. This can be done with the following code:

import numpy as np
import pandas as pd
import tensorflow as tf

human_labels_np_path = '<local_path>/CIFAR-100_human_ordered.npy'
human_labels_csv_path = '<local_path>/CIFAR-100_human_annotations.csv'

with, "rb") as f:
  human_annotations = np.load(f, allow_pickle=True)

df = pd.DataFrame(human_annotations[()])

with, "w") as f:
  df.to_csv(f, index=False)
Split Examples
'test' 10,000
'train' 50,000
  • Feature structure:
    'coarse_label': ClassLabel(shape=(), dtype=tf.int64, num_classes=20),
    'id': Text(shape=(), dtype=tf.string),
    'image': Image(shape=(32, 32, 3), dtype=tf.uint8),
    'label': ClassLabel(shape=(), dtype=tf.int64, num_classes=100),
    'noise_label': ClassLabel(shape=(), dtype=tf.int64, num_classes=100),
    'worker_id': tf.int64,
    'worker_time': tf.float32,
  • Feature documentation:
Feature Class Shape Dtype Description
coarse_label ClassLabel tf.int64
id Text tf.string
image Image (32, 32, 3) tf.uint8
label ClassLabel tf.int64
noise_label ClassLabel tf.int64
worker_id Tensor tf.int64
worker_time Tensor tf.float32


  • Citation:
  title={Learning with Noisy Labels Revisited: A Study Using Real-World Human
  author={Jiaheng Wei and Zhaowei Zhu and Hao Cheng and Tongliang Liu and Gang
  Niu and Yang Liu},
  booktitle={International Conference on Learning Representations},