HackerPapers

arXiv:1711.11248cs.CV

A Closer Look at Spatiotemporal Convolutions for Action Recognition

Du Tran, Heng Wang, Lorenzo Torresani, Jamie Ray, Yann LeCun, Manohar Paluri

HN submission. Found in a submission link or body. Submitters do not count as commenters in the main ranking; a submission is not an endorsement.

Abstract

In this paper we discuss several forms of spatiotemporal convolutions for video analysis and study their effects on action recognition. Our motivation stems from the observation that 2D CNNs applied to individual frames of the video have remained solid performers in action recognition. In this work we empirically demonstrate the accuracy advantages of 3D CNNs over 2D CNNs within the framework of residual learning. Furthermore, we show that factorizing the 3D convolutional filters into separate spatial and temporal components yields significantly advantages in accuracy. Our empirical study leads to the design of a new spatiotemporal convolutional block "R(2+1)D" which gives rise to CNNs that achieve results comparable or superior to the state-of-the-art on Sports-1M, Kinetics, UCF101 and HMDB51.

arXiv comments
Journal reference
DOI
Subjects
cs.CVComputer Vision and Pattern Recognition

On Hacker News

No comments linking this paper were found in the collected sources. Found in 1 HN submission from Sep 2019 to Sep 2019. Open submission discussions.

Comments linking this paper, per year.

Discussed in comments0 threads

No comments linking this paper were found in the collected sources. Open submission discussions.

Submitted to Hacker News

  • IG65M-PyTorch: video models pre-trained on over 65M Instagram videos10 points0 comments

    Think: resnet+imagenet but for videos: https://github.com/moabitcoin/ig65m-pytorch We ported the r(2+1)d model (from CVPR 2018, see https://arxiv.org/abs/1711.11248) and the weights pre-trained by Facebook Research on over 65 million Instagram videos (from CVPR 2019, https://arxiv.org/abs/1905.00561) to PyTorch and released architecture, weights, tools for conversion, and feature extraction example. The official Facebook Research codebase can be found at https://github.com/facebookresearch/vmz…

Connected through shared HN threads, not inferred similarity or a recommendation.