Datasets:

pzhou10
/

3D-PC

split stringclasses 1 value	label float64 0 1	category stringclasses 10 values	scene stringclasses 19 values	setting int64 1 11	img_id stringclasses 50 values
train	1	bench	bench_01	1	0000.png
train	1	bench	bench_01	1	0001.png
train	1	bench	bench_01	1	0002.png
train	1	bench	bench_01	1	0003.png
train	1	bench	bench_01	1	0004.png
train	1	bench	bench_01	1	0005.png
train	1	bench	bench_01	1	0006.png
train	1	bench	bench_01	1	0007.png
train	1	bench	bench_01	1	0008.png
train	1	bench	bench_01	1	0009.png
train	1	bench	bench_01	1	0012.png
train	1	bench	bench_01	1	0013.png
train	1	bench	bench_01	1	0014.png
train	1	bench	bench_01	1	0015.png
train	1	bench	bench_01	1	0017.png
train	1	bench	bench_01	1	0018.png
train	0	bench	bench_01	1	0024.png
train	0	bench	bench_01	1	0039.png
train	0	bench	bench_01	2	0015.png
train	0	bench	bench_01	2	0016.png
train	0	bench	bench_01	2	0017.png
train	0	bench	bench_01	2	0018.png
train	0	bench	bench_01	2	0019.png
train	0	bench	bench_01	2	0020.png
train	0	bench	bench_01	2	0022.png
train	0	bench	bench_01	2	0023.png
train	1	bench	bench_01	2	0031.png
train	1	bench	bench_01	2	0034.png
train	1	bench	bench_01	2	0035.png
train	1	bench	bench_01	2	0036.png
train	1	bench	bench_01	2	0037.png
train	1	bench	bench_01	2	0038.png
train	1	bench	bench_01	2	0039.png
train	1	bench	bench_01	2	0040.png
train	1	bench	bench_01	2	0041.png
train	1	bench	bench_01	2	0042.png
train	1	bench	bench_01	2	0043.png
train	1	bench	bench_01	2	0044.png
train	1	bench	bench_01	2	0045.png
train	1	bench	bench_01	2	0046.png
train	1	bench	bench_01	2	0047.png
train	1	bench	bench_01	2	0048.png
train	1	bench	bench_01	2	0049.png
train	1	bench	bench_01	3	0000.png
train	1	bench	bench_01	3	0001.png
train	1	bench	bench_01	3	0002.png
train	1	bench	bench_01	3	0009.png
train	1	bench	bench_01	3	0010.png
train	1	bench	bench_01	3	0011.png
train	1	bench	bench_01	3	0012.png
train	1	bench	bench_01	3	0013.png
train	1	bench	bench_01	3	0014.png
train	0	bench	bench_01	3	0020.png
train	0	bench	bench_01	3	0021.png
train	0	bench	bench_01	3	0022.png
train	0	bench	bench_01	3	0024.png
train	0	bench	bench_01	3	0039.png
train	1	bench	bench_01	3	0048.png
train	1	bench	bench_01	3	0049.png
train	1	bench	bench_01	4	0010.png
train	1	bench	bench_01	4	0011.png
train	1	bench	bench_01	4	0012.png
train	1	bench	bench_01	4	0013.png
train	1	bench	bench_01	4	0014.png
train	1	bench	bench_01	4	0015.png
train	1	bench	bench_01	4	0016.png
train	1	bench	bench_01	4	0023.png
train	0	bench	bench_01	4	0037.png
train	0	bench	bench_01	4	0038.png
train	0	bench	bench_01	4	0040.png
train	0	bench	bench_01	4	0041.png
train	0	bench	bench_01	4	0042.png
train	0	bench	bench_01	4	0043.png
train	0	bench	bench_01	4	0044.png
train	0	bench	bench_01	4	0045.png
train	0	bench	bench_01	4	0046.png
train	0	bench	bench_01	4	0047.png
train	0	bench	bench_01	4	0048.png
train	0	bench	bench_01	4	0049.png
train	1	bench	bench_01	5	0001.png
train	1	bench	bench_01	5	0003.png
train	1	bench	bench_01	5	0004.png
train	1	bench	bench_01	5	0005.png
train	1	bench	bench_01	5	0006.png
train	1	bench	bench_01	5	0007.png
train	1	bench	bench_01	5	0008.png
train	1	bench	bench_01	5	0009.png
train	1	bench	bench_01	5	0010.png
train	1	bench	bench_01	5	0011.png
train	1	bench	bench_01	5	0012.png
train	1	bench	bench_01	5	0013.png
train	1	bench	bench_01	5	0014.png
train	1	bench	bench_01	5	0015.png
train	1	bench	bench_01	5	0016.png
train	1	bench	bench_01	5	0017.png
train	0	bench	bench_01	5	0038.png
train	0	bench	bench_01	5	0039.png
train	0	bench	bench_01	5	0040.png
train	0	bench	bench_01	6	0000.png
train	0	bench	bench_01	6	0001.png

End of preview. Expand in Dataset Viewer.

Project Description

arXiv: https://arxiv.org/abs/2406.04138
GitHub Repo: https://github.com/serre-lab/VPT
Project Page: https://serre-lab.github.io/VPT

Summary

Visual perspective taking (VPT), the ability to accurately perceive and reason about the perspectives of others, is an essential feature of human intelligence. Deep neural networks (DNNs) may be a good candidate for modeling VPT and its computational demands in light of a growing number of reports indicating that DNNs gain the ability to analyze 3D scenes after training on large static-image datasets. We developed the 3D perception challenge (3D-PC) for comparing 3D perceptual capabilities in humans and DNNs. The 3D-PC is comprised of three 3D-analysis tasks posed within natural scene images:

A simple test of object depth order (depth),
A basic VPT task (vpt-basic),
A version of VPT (vpt-strategy) designed to limit the effectiveness of "shortcut" visual strategies.

Citation

@misc{linsley20243dpc,
      title={The 3D-PC: a benchmark for visual perspective taking in humans and machines}, 
      author={Drew Linsley and Peisen Zhou and Alekh Karkada Ashok and Akash Nagaraj and Gaurav Gaonkar and Francis E Lewis and Zygmunt Pizlo and Thomas Serre},
      year={2024},
      eprint={2406.04138},
      archivePrefix={arXiv},
      primaryClass={cs.CV}
}

Downloads last month: 5

Edit dataset card

Size of downloaded dataset files:

17.5 GB

Size of the auto-converted Parquet files:

17.5 GB

Number of rows:

11,862