Tiny ImageNet

Overview

Tiny ImageNet is a downsampled subset of the ImageNet ILSVRC dataset created for the Tiny ImageNet Visual Recognition Challenge. It contains 200 classes and approximately 100,000 images resized to 64×64 pixels in RGB format. For each class there are typically 500 training images and 50 validation images.

• Train: ~100,000 images total (500 images per class)

• Validation: 10,000 images (50 images per class)

Data Structure

When accessing an example using ds[i] you will receive a dictionary with these keys:

Table 1 :header-rows: 1 :widths: 20 20 60

Key	Type	Description
image	PIL.Image.Image	64×64×3 RGB image
label	int	Class label (0-199) or string class id depending on dataset builder

Usage Example

Basic Usage

from stable_datasets.images.tiny_imagenet import TinyImagenet

# First run will download + prepare cache, then return the split as a HF Dataset
ds = TinyImagenet(split="train")

# If you omit the split (split=None), you get a DatasetDict with all available splits
ds_all = TinyImagenet(split=None)

sample = ds[0]
print(sample.keys())  # {"image", "label"}

# Optional: make it PyTorch-friendly
ds_torch = ds.with_format("torch")

Related Datasets

• tiny_imagenet_c : Corrupted version with common corruptions applied

• CIFAR variants: smaller benchmarks with similar tasks (see cifar10, cifar100)

References

• Tiny ImageNet (CS231n page): http://cs231n.stanford.edu/tiny-imagenet-200.html

• Competition / resources: https://www.kaggle.com/c/tiny-imagenet

Citation

@inproceedings{Le2015TinyIV,
        title={Tiny ImageNet Visual Recognition Challenge},
        author={Ya Le and Xuan S. Yang},
        year={2015}
}