Pic 01

Optimization
and priors

We are specially focussed on soliciting papers on dense monocular reconstruction.

Pic 02

Semantic
elements

We are also looking for papers on new methods for integrating semantics to make reconstructions more robust.

Pic 02

Efficiency
and compression

And new methods for making reconstruction more efficient and suitable for realtime application.

Call for papers

We are soliciting high quality papers covering, but not limited to, the topics listed below. Paper length should be up to 8 pages.
Papers should follow the standard CVPR formatting instructions. Accepted papers will appear in the CVPR workshop proceedings.
Submission is via CMT.
Submission deadline: April 12, 2018, Anywhere on Earth (UTC-12) (Hard deadline)
Author notification: April 18, 2018
Camera ready: April 20, 2018

Topics

  • Deep learning pipelines for Visual SLAM
  • Uncertaininty in Deep Networks for Integration with traditional systems
  • Supervised, Unsupervised and Adversarial Learning for Real-time Odometry, Tracking and Ego-Motion estimation
  • Supervised, Unsupervised and Adversarial Learning for Single and Multi-View 3D Reconstruction
  • Meta-optimizers and learnt priors for visual SLAM
  • Spatial Memory-Augmented Deep Networks
  • Generative Models for Place recognition and Relocalization
  • Semantics for SLAM
  • New methods for 3D Scene Representation and Compression
  • Deep Learning and SLAM Efficiency