Abstract
We propose a method for large displacement optical flow in
which local matching costs are learned by a convolutional neural network
(CNN) and a smoothness prior is imposed by a conditional random
field (CRF). We tackle the computation- and memory-intensive operations
on the 4D cost volume by a min-projection which reduces memory
complexity from quadratic to linear and binary descriptors for efficient
matching. This enables evaluation of the cost on the fly and allows to
perform learning and CRF inference on high resolution images without
ever storing the 4D cost volume. To address the problem of learning binary
descriptors we propose a new hybrid learning scheme. In contrast
to current state of the art approaches for learning binary CNNs we can
compute the exact non-zero gradient within our model. We compare several
methods for training binary descriptors and show results on public
available benchmarks.
which local matching costs are learned by a convolutional neural network
(CNN) and a smoothness prior is imposed by a conditional random
field (CRF). We tackle the computation- and memory-intensive operations
on the 4D cost volume by a min-projection which reduces memory
complexity from quadratic to linear and binary descriptors for efficient
matching. This enables evaluation of the cost on the fly and allows to
perform learning and CRF inference on high resolution images without
ever storing the 4D cost volume. To address the problem of learning binary
descriptors we propose a new hybrid learning scheme. In contrast
to current state of the art approaches for learning binary CNNs we can
compute the exact non-zero gradient within our model. We compare several
methods for training binary descriptors and show results on public
available benchmarks.
Original language | English |
---|---|
Publication status | Published - 13 Sept 2017 |
Event | 39th German Conference on Pattern Recognition - Basel, Switzerland Duration: 13 Sept 2016 → 15 Sept 2017 https://gcpr2017.dmi.unibas.ch/en/ |
Conference
Conference | 39th German Conference on Pattern Recognition |
---|---|
Abbreviated title | GCPR 2017 |
Country/Territory | Switzerland |
City | Basel |
Period | 13/09/16 → 15/09/17 |
Internet address |
Fingerprint
Dive into the research topics of 'Scalable Full Flow with Learned Binary Descriptors'. Together they form a unique fingerprint.Prizes
-
GCPR Honorable Mention
Munda, G. (Recipient), Shekhovtsov, O. (Recipient), Knöbelreiter, P. (Recipient) & Pock, T. (Recipient), 15 Sept 2017
Prize: Prizes / Medals / Awards