Unsupervised Learning of Spatiotemporally Coherent Metrics
Ross Goroshin,Joan Bruna,Jonathan Tompson,David Eigen,Yann LeCun
(Submitted on 18 Dec 2014 (v1), last revised 8 Sep 2015 (this version, v6))
Current state-of-the-art classification and detection algorithms rely on supervised training. In this work we study unsupervised feature learning in the context of temporally coherent video data. We focus on feature learning from unlabeled video data, using the assumption that adjacent video frames contain semantically similar information. This assumption is exploited to train a convolutional pooling auto-encoder regularized by slowness and sparsity. We establish a connection between slow feature learning to metric learning and show that the trained encoder can be used to define a more temporally and semantically coherent metric.
Comments:To appear at ICCV2015
Subjects:Computer Vision and Pattern Recognition (cs.CV)
Cite as:arXiv:1412.6056[cs.CV]
(orarXiv:1412.6056v6[cs.CV]for this version)
Submission history
From: Rostislav Goroshin [view email]
[v1]Thu, 18 Dec 2014 20:31:56 GMT (3179kb,D)
[v2]Tue, 23 Dec 2014 21:57:59 GMT (3391kb,D)
[v3]Sat, 27 Dec 2014 17:16:47 GMT (3391kb,D)
[v4]Thu, 26 Feb 2015 19:28:21 GMT (4505kb,D)
[v5]Fri, 4 Sep 2015 19:46:15 GMT (4462kb,D)
[v6]Tue, 8 Sep 2015 18:39:03 GMT (4462kb,D)
Download:
