v0.5.0

Open3D 0.5.0 Release Notes

The Open3D team and the Open Source Vision Foundation (http://www.osvf.org) are excited to announce the 0.5.0 release of the Open3D library.

In this release we show the power of Open3D as a core tool to create machine learning solutions for 3D data.

We introduce a re-implementation of the PointNet++ architecture to perform point cloud semantic segmentation using Open3D and TensorFlow. Our Open3D-PointNet++ is able to produce highly accurate results in the Semantic3D benchmark, surpassing the results of the original PointNet++ implementation. Even more exciting is the fact that our re-designed Open3D-PointNet++ is able to perform real-time inference (+10 FPS) on the KITTI dataset. We show how to perform training and inference of Open3D-PointNet++ in both Semantic3D and KITTI. Check out this blog post for more information!

We have also added a new VoxelGrid representation and tooling to convert from point clouds to a VoxelGrid structure. This functionality is extremely useful to produce representations that are easier to digest by neural networks.

We have also done significant improvements to our internal infrastructure, including a simplified CI testing mechanism via docker images, enhanced testing coverage, and easier installation of the library.

Find the full change log here.

Enjoy!
The Open3D team