Elriggs/imagenet-50-subset

ImageNet-50 Subset

This dataset contains the first 50 classes from ImageNet-1K with up to 1,000 images per class (where available).

Dataset Statistics

• Total Classes: 50
• Total Images: 50000
• Train/Val Split: 90%/10%
• Max Images per Class: 1000

Dataset Structure

imagenet-50-subset/
├── train/
│   ├── n01440764/  # tench
│   │   ├── n01440764_1234.JPEG
│   │   └── ...
│   ├── n01443537/  # goldfish
│   └── ...
├── val/
│   ├── n01440764/
│   ├── n01443537/
│   └── ...
├── metadata.json
├── wnid_to_class.txt
└── README.md

Classes Included

WordNet ID	Class Name	Train Images	Val Images	Total
n01440764	tench	900	100	1000
n01443537	goldfish	900	100	1000
n01484850	great white shark	900	100	1000
n01491361	tiger shark	900	100	1000
n01494475	hammerhead	900	100	1000
n01496331	electric ray	900	100	1000
n01498041	stingray	900	100	1000
n01514668	cock	900	100	1000
n01514859	hen	900	100	1000
n01518878	ostrich	900	100	1000
n01530575	brambling	900	100	1000
n01531178	goldfinch	900	100	1000
n01532829	house finch	900	100	1000
n01534433	junco	900	100	1000
n01537544	indigo bunting	900	100	1000
n01558993	robin	900	100	1000
n01560419	bulbul	900	100	1000
n01580077	jay	900	100	1000
n01582220	magpie	900	100	1000
n01592084	chickadee	900	100	1000
n01601694	water ouzel	900	100	1000
n01608432	kite	900	100	1000
n01614925	bald eagle	900	100	1000
n01616318	vulture	900	100	1000
n01622779	great grey owl	900	100	1000
n01629819	European fire salamander	900	100	1000
n01630670	common newt	900	100	1000
n01631663	eft	900	100	1000
n01632458	spotted salamander	900	100	1000
n01632777	axolotl	900	100	1000
n01641577	bullfrog	900	100	1000
n01644373	tree frog	900	100	1000
n01644900	tailed frog	900	100	1000
n01664065	loggerhead	900	100	1000
n01665541	leatherback turtle	900	100	1000
n01667114	mud turtle	900	100	1000
n01667778	terrapin	900	100	1000
n01669191	box turtle	900	100	1000
n01675722	banded gecko	900	100	1000
n01677366	common iguana	900	100	1000
n01682714	American chameleon	900	100	1000
n01685808	whiptail	900	100	1000
n01687978	agama	900	100	1000
n01688243	frilled lizard	900	100	1000
n01689811	alligator lizard	900	100	1000
n01692333	Gila monster	900	100	1000
n01693334	green lizard	900	100	1000
n01694178	African chameleon	900	100	1000
n01695060	Komodo dragon	900	100	1000
n01697457	African crocodile	900	100	1000

Usage with Hugging Face Datasets

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("your-username/imagenet-50-subset")

# Access train and validation splits
train_dataset = dataset['train']
val_dataset = dataset['validation']

# Example: Load and display an image
from PIL import Image
import matplotlib.pyplot as plt

sample = train_dataset[0]
image = Image.open(sample['image'])
label = sample['label']
plt.imshow(image)
plt.title(f"Class: {label}")
plt.show()

Usage with PyTorch

from torchvision import datasets, transforms
from torch.utils.data import DataLoader

# Define transforms
transform = transforms.Compose([
    transforms.Resize(256),
    transforms.CenterCrop(224),
    transforms.ToTensor(),
    transforms.Normalize(mean=[0.485, 0.456, 0.406],
                       std=[0.229, 0.224, 0.225])
])

# Load datasets
train_dataset = datasets.ImageFolder('./imagenet-50-subset/train', transform=transform)
val_dataset = datasets.ImageFolder('./imagenet-50-subset/val', transform=transform)

# Create data loaders
train_loader = DataLoader(train_dataset, batch_size=32, shuffle=True)
val_loader = DataLoader(val_dataset, batch_size=32, shuffle=False)

License

This subset inherits the ImageNet license. Please ensure you have the right to use ImageNet data. The original ImageNet dataset is available at http://www.image-net.org/

Citation

If you use this dataset, please cite the original ImageNet paper:

@article{deng2009imagenet,
  title={Imagenet: A large-scale hierarchical image database},
  author={Deng, Jia and Dong, Wei and Socher, Richard and Li, Li-Jia and Li, Kai and Fei-Fei, Li},
  journal={2009 IEEE conference on computer vision and pattern recognition},
  pages={248--255},
  year={2009},
  organization={IEEE}
}