You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We are implementing an amazing feature for Open3D that trying to support Paddle backend for Open3D ml module.
What is a Paddle ?
PaddlePaddle (Parallel Distributed Deep Learning) is an open-source deep learning platform originally developed by Baidu. It is designed to facilitate the development and deployment of deep learning models with support for a wide range of applications, from natural language processing (NLP) to computer vision and speech recognition. PaddlePaddle offers an easy-to-use interface, extensive libraries, and robust ecosystem tools, making it suitable for both beginners and experts in the field of deep learning.
Why is Open3D ?
The Paddle community is committed to expanding its capabilities in the field of scientific computing. Open3D, as a feature-rich library for 3D data processing, offers functionalities such as point cloud registration, geometric transformations, and 3D visualization. It is widely used in various fields, including computer vision, robotics, medical imaging, and materials science. Paddle adding support for Open3Dcan further enhance its capabilities in scientific computing.
Target Platform
Our work will start on linux x86 and we have no plan to support other arch or os currently.
RoadMap
For open3d.ml.paddle the layer API should be aligned to open3d.ml.pytorch but functional APi is hard to align due to limitation of Paddle C++ operators.
We divided our work into 3 parts. Following is roadmap :
Part 1
Integrate Paddle backend to Open3D build.
Implement total layer in neighbor_search.py and corresponding C++ operators.
Checklist
main
branch).Proposed new feature or change
Description
We are implementing an amazing feature for Open3D that trying to support Paddle backend for Open3D ml module.
What is a Paddle ?
PaddlePaddle (Parallel Distributed Deep Learning) is an open-source deep learning platform originally developed by Baidu. It is designed to facilitate the development and deployment of deep learning models with support for a wide range of applications, from natural language processing (NLP) to computer vision and speech recognition. PaddlePaddle offers an easy-to-use interface, extensive libraries, and robust ecosystem tools, making it suitable for both beginners and experts in the field of deep learning.
Why is Open3D ?
The Paddle community is committed to expanding its capabilities in the field of scientific computing. Open3D, as a feature-rich library for 3D data processing, offers functionalities such as point cloud registration, geometric transformations, and 3D visualization. It is widely used in various fields, including computer vision, robotics, medical imaging, and materials science. Paddle adding support for Open3Dcan further enhance its capabilities in scientific computing.
Target Platform
Our work will start on linux x86 and we have no plan to support other arch or os currently.
RoadMap
For
open3d.ml.paddle
the layer API should be aligned toopen3d.ml.pytorch
but functional APi is hard to align due to limitation of Paddle C++ operators.We divided our work into 3 parts. Following is roadmap :
Part 1
neighbor_search.py
and corresponding C++ operators.PR: #7028
Part 2
Part 3
Influence
Supporting Paddle backend for Open3D will not influence existing functions.
References
No response
Additional information
No response
The text was updated successfully, but these errors were encountered: