zlm200/some-implementions-with-rgbd-camera-realsense-d435

some personal implementation of 3d object reconstruction, recognition and pose estimation

This repository may be no longer maintained.

---

3D Reconstruction with RealSenseD435

Some personal implementation of 3d reconstruction with Realsense D435.

Contents

• Basic

1.capture RGBD pointcloud and save RGBD images

2.Record RGBD stream as '.bag' file

3.Read RGBD '.bag' file

• 3D Object Reconstruction Based on Muti-view RGBD Images Using An A4 paper

Using an A4 paper to reconstruct an object's 3d pointcloud.

• Realtime 3D Reconstruction Using Double Opposing RealSenses

Using a double side chessBoard to realtime registrating pointclouds from two opposite rgbd camera.

• Object Recognition Using PointNet

Realtime 3d object partial pointcloud recognition based on PointNet.

• Object Pose Estimation Using DenseFusion

Realtime object pose estimation using DenseFusion.

Screenshots

point cloud of a toy dog

realtime registration of pointclouds from two cams.

Realtime object recognition using PointNet.

Realtime object pose estimation using DenseFusion.

(Using the box's pose to control the virtual dog model's pose)