English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Learning-based visual odometry and SLAM methods demonstrate a steadyimprovement over past years. However, collecting ground truth poses to trainthese methods is difficult and expensive. This could be resolved by training inan unsupervised mode, but there is still a large gap between performance ofunsupervised and supervised methods. In this work, we focus on generatingsynthetic data for deep learning-based visual odometry and SLAM methods thattake optical flow as an input. We produce training data in a form of opticalflow that corresponds to arbitrary camera movement between a real frame and avirtual frame. For synthesizing data we use depth maps either produced by adepth sensor or estimated from stereo pair. We train visual odometry model onsynthetic data and do not use ground truth poses hence this model can beconsidered unsupervised. Also it can be classified as monocular as we do notuse depth maps on inference. We also propose a simple way to convert any visualodometry model into a SLAM method based on frame matching and graphoptimization. We demonstrate that both the synthetically-trained visualodometry model and the proposed SLAM method build upon this model yieldsstate-of-the-art results among unsupervised methods on KITTI dataset and showspromising results on a challenging EuRoC dataset.

Training Deep SLAM on Single Frames