vrachit/imagenet-1k-webdataset
收藏Hugging Face2025-12-26 更新2026-03-29 收录
下载链接:
https://hf-mirror.com/datasets/vrachit/imagenet-1k-webdataset
下载链接
链接失效反馈官方服务:
资源简介:
# ImageNet-1k WebDataset
This dataset contains ImageNet-1k in WebDataset format (tar files) for efficient streaming.
## Dataset Structure
- **Training**: 129 shards (train-*.tar)
- **Validation**: 5 shards (validation-*.tar)
- **Total size**: 147.82 GB
## Format
Each tar file contains samples with:
- `*.jpg`: Image bytes
- `*.cls`: Label (class ID as text)
## Usage
```python
import webdataset as wds
# Training dataset
train_url = "train-{000000..000000000}.tar"
dataset = wds.WebDataset(train_url).shuffle(1000).decode("rgb")
for sample in dataset:
image = sample["jpg"] # PIL Image
label = int(sample["cls"])
# ... your training code
```
## Source
Converted from ILSVRC/imagenet-1k parquet format.
Generated using [MetaDistil ImageNet Pipeline](https://modal.com)
提供机构:
vrachit



