search

F2Wang / ObjectDatasetTools

Tools to create pixel-wise object masks, bounding box labels (2D and 3D) and 3D object model (PLY triangle mesh) for object sequences filmed with an RGB-D camera. This project prepares training and testing data for various deep learning projects such as 6D object pose estimation projects singleshotpose, as well as object detection and instance segmentation projects.
MIT License
411 stars 90 forks source link

example sequence - timer result #41

Open vmvz opened 4 years ago

vmvz commented 4 years ago

The result is wrong. What did I do wrong? thanks!

Step:

  1. install env (Ubuntu 18.04 Python3.6)
  2. download example sequence (timer)
  3. python3 order 
    python compute_gt_poses.py LINEMOD/timer
python register_segmented.py LINEMOD/timer
python create_label_files.py LINEMOD/timer
  4. result: timer.ply timer.ply.zip

image

F2Wang commented 4 years ago

Thank you for bringing up the issue. This was my fault, I added a line to test the new API and forgot to delete it. This line will force the algorithm to ignore all information provided the marker and performs matching based on ICP only, which caused the misalignment shown above and make the algorithm super slow.

Also please note that register_segmented only work sometimes, if auto meshing fails, you can either run register_Segmented with meshing option turned off, and remove noise and perform meshing in meshlab. Or you can use register_scene and remove all noise manually.

Please pull the latest change and let me know if that helps. Thanks