TY - JOUR
T1 - Rel3D
T2 - 34th Conference on Neural Information Processing Systems, NeurIPS 2020
AU - Goyal, Ankit
AU - Yang, Kaiyu
AU - Yang, Dawei
AU - Deng, Jia
N1 - Funding Information:
Acknowledgement. This work is partially supported by the National Science Foundation under Grant IIS-1734266 and the Office of Naval Research under Grant N00014-20-1-2634. We would like to thank Alexander Strzalkowski, Pranay Manocha and Shruti Bhargava for their help with data collection.
Publisher Copyright:
© 2020 Neural information processing systems foundation. All rights reserved.
PY - 2020
Y1 - 2020
N2 - Understanding spatial relations (e.g., “laptop on table”) in visual input is important for both humans and robots. Existing datasets are insufficient as they lack large-scale, high-quality 3D ground truth information, which is critical for learning spatial relations. In this paper, we fill this gap by constructing Rel3D: the first large-scale, human-annotated dataset for grounding spatial relations in 3D. Rel3D enables quantifying the effectiveness of 3D information in predicting spatial relations on large-scale human data. Moreover, we propose minimally contrastive data collection—a novel crowdsourcing method for reducing dataset bias. The 3D scenes in our dataset come in minimally contrastive pairs: two scenes in a pair are almost identical, but a spatial relation holds in one and fails in the other. We empirically validate that minimally contrastive examples can diagnose issues with current relation detection models as well as lead to sample-efficient training. Code and data are available at https://github.com/princeton-vl/Rel3D.
AB - Understanding spatial relations (e.g., “laptop on table”) in visual input is important for both humans and robots. Existing datasets are insufficient as they lack large-scale, high-quality 3D ground truth information, which is critical for learning spatial relations. In this paper, we fill this gap by constructing Rel3D: the first large-scale, human-annotated dataset for grounding spatial relations in 3D. Rel3D enables quantifying the effectiveness of 3D information in predicting spatial relations on large-scale human data. Moreover, we propose minimally contrastive data collection—a novel crowdsourcing method for reducing dataset bias. The 3D scenes in our dataset come in minimally contrastive pairs: two scenes in a pair are almost identical, but a spatial relation holds in one and fails in the other. We empirically validate that minimally contrastive examples can diagnose issues with current relation detection models as well as lead to sample-efficient training. Code and data are available at https://github.com/princeton-vl/Rel3D.
UR - http://www.scopus.com/inward/record.url?scp=85106640086&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85106640086&partnerID=8YFLogxK
M3 - Conference article
AN - SCOPUS:85106640086
SN - 1049-5258
VL - 2020-December
JO - Advances in Neural Information Processing Systems
JF - Advances in Neural Information Processing Systems
Y2 - 6 December 2020 through 12 December 2020
ER -