Skip to content

CVPR 2020 release

Compare
Choose a tag to compare
@AdrienGaidon-TRI AdrienGaidon-TRI released this 10 Jun 17:01
· 7 commits to master since this release
dfbdc27

Complete set of features described in our CVPR 2020 oral: 3D Packing for Self-Supervised Monocular Depth Estimation.

Additions from v0.1.0:

  • Velocity loss (see VelSupModel)
  • PackNetSlim01 (faster version of PackNet)
  • Support for multi-camera loading on the DGP dataset
  • Support for fp16 at inference time