Pixel-ImageNet

The subset of the dataset is available. We will release the remaining part of the dataset later. Please stay tuned.

We updated the subset of Pixel-ImageNet. The newly released version has 0.06M more images than before.

Introduction

The Pixel-ImageNet contains 1000 classes and 0.615M instance masks of ILSVRCLOC collected using IOG. In the Pixel-ImageNet, each image contains only one class and the pixel value in the groundtruth is used to distinguish each instance.

Download

This is the subset of Pixel-ImageNet, which contains 0.485M images and spans 946 object classes. The detail of the object class and the number of images per class is shown in label-v2.npy. The dataset is available on Google Drive and Baidu Netdisk (pwd: i1u0). This subset is based on the training set of ImageNet. The full Pixel-ImageNet dataset will be released later. Please stay tuned.

Citations

Please consider citing our dataset in your publications if it helps your research. The following is a BibTeX reference.

@inproceedings{zhang2020interactive,
  title={Interactive Object Segmentation With Inside-Outside Guidance},
  author={Zhang, Shiyin and Liew, Jun Hao and Wei, Yunchao and Wei, Shikui and Zhao, Yao},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={12234--12244},
  year={2020}
}