Scalable Full Flow with Learned Binary Descriptors

Gottfried Munda, Alexander Shekhovtsov, Patrick Knöbelreiter, Thomas Pock

Publikation: KonferenzbeitragPaperBegutachtung


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.
PublikationsstatusVeröffentlicht - 13 Sep 2017
Veranstaltung39th German Conference on Pattern Recognition - Basel, Schweiz
Dauer: 13 Sep 201615 Sep 2017


Konferenz39th German Conference on Pattern Recognition
Kurztitel GCPR 2017


Untersuchen Sie die Forschungsthemen von „Scalable Full Flow with Learned Binary Descriptors“. Zusammen bilden sie einen einzigartigen Fingerprint.
  • GCPR Honorable Mention

    Munda, G. (Empfänger/-in), Shekhovtsov, O. (Empfänger/-in), Knöbelreiter, P. (Empfänger/-in) & Pock, Thomas (Empfänger/-in), 15 Sep 2017

    Auszeichnung: Preise / Medaillen / Ehrungen

Dieses zitieren