PASS: Pictures without humAns for Self-Supervised Pretraining
TL;DR: An ImageNet replacement dataset for self-supervised pretraining without humans
Content
PASS is a large-scale image dataset that does not include any humans, human parts, or other personally identifiable information that can be used for high-quality pretraining while significantly reducing privacy concerns.
Download the dataset
Generally: all information is on our webpage.
For downloading the dataset, please visit our dataset on zenodo. There you can download it in tar files and find the meta-data.
You can also download the images from their AWS urls, from here.
Pretrained models
| Pretraining | Method | Epochs | Places205 lin. Acc. | Model weights |
|---|---|---|---|---|
| IN-1k | MoCo-v2 | 200 | 50.1 | R50 weights |
| PASS | MoCo-v2 | 200 | 52.8 | R50 weights |
| PASS | MoCo-v2-CLD | 200 | 53.1 | R50 weights |
| PASS | SwAV | 200 | 55.5 | R50 weights |
| PASS | DINO | 100 | X | ViT S16 weights |
| PASS | DINO | 300 | coming soon | |
| PASS | MoCo-v2 | 800 | coming soon |
Contribute your models
Please let us know if you have a model pretrained on this dataset and I will add this to the list above.
Citation
@Article{asano21pass,
author = "Yuki M. Asano and Christian Rupprecht and Andrew Zisserman and Andrea Vedaldi",
title = "PASS: An ImageNet replacement for self-supervised pretraining without humans",
journal = "NeurIPS Track on Datasets and Benchmarks",
year = "2021"
}
