Home

Awesome

CloudAAE

This is an tensorflow implementation of "CloudAAE: Learning 6D Object Pose Regression with On-line Data Synthesis on Point Clouds"

Files

  1. log: directory to store log files during training.
  2. losses: loss functions for training.
  3. models: a python file defining model structure.
  4. object_model_tfrecord: full object models for data synthesizing and visualization purpose.
  5. tf_ops: tensorflow implementation of sampling operations (credit: Haoqiang Fan, Charles R. Qi).
  6. trained_network: a trained network.
  7. utils: utility files for defining model structure.
  8. ycb_video_data_tfRecords: synthetic training data and real test data for the YCB video dataset.
  9. evaluate_cloudAAE_ycbv.py: script for testing object 6d pose estimation with a trained network on test set in YCB video dataset.
  10. train_cloudAAE_ycbv.py: script for training a network on synthetic data for YCB objects.

Requirements

Test a trained network

  1. Testing data in tfrecord format is available
  1. After activate tensorflow
python evaluate_cloudAAE_ycbv.py --trained_model trained_network/20200908-204328/model.ckpt --batch_size 1 --target_cls 0
  1. Result
<p float="center"> <img src="/figure/0.gif" width="200" /> <img src="/figure/0_recon.gif" width="200" /> <img src="/figure/14.gif" width="200" /> <img src="/figure/14_recon.gif" width="200" /> </p>

Train a network

  1. Training data is created synthetically using 3D object model and 6D poses.
<p float="center"> <img src="/figure/syn_and_real_seg_lamp.gif" width="400" /> <img src="/figure/syn_and_real_seg_phone.gif" width="400" /> </p>
  1. Run script
python train_cloudAAE_ycbv.py
  1. Log files and trained model is store in log

Citation

If you use this code in an academic context, please consider cite the paper:

BiBTeX:

@inproceedings{gao2020cloudpose,
      title={CloudAAE: Learning 6D Object Pose Regression with On-line Data
Synthesis on Point Clouds},
      author={G. Gao, M. Lauri, X. Hu, J. Zhang and S. Frintrop},
      booktitle={ICRA},
      year={2021}
    }

Link to Paper

TBA

Acknowledgement