Turntable 3D Reconstruction for WAAM Parts

This project focuses on 3D reconstruction for post-processing of Wire Arc Additive Manufacturing (WAAM) parts. Below is a comparison between the reconstructed 3D model and the actual object:

Reconstructed Model	Actual Object

Workflow Overview

The reconstruction process is carried out using the following scripts in this folder:

1. aruco.bat
   Performs turntable-camera calibration using ZED SDK and outputs a transforms.json to unproject points from camera frame to world frame.

2. run.bat
   Collects RGB and Depth data using ZED SDK and converts to the format required for segmentation and reconstruction.

3. find_box_point.py
   Can be used to define bounding box and positive/negative selection points for segmentation using SAM2.

4. sam2_segmentation.py
   Segments and saves output mask of object.

5. s3d_reconstruction.py
   Reconstructs object using object masks, turntable transforms, and depth maps.

Performance Evaluation

• Precision: 1.9 - 2 cms
• Processing Time: 20 - 30 secs