CloudPose

KO | EN

gitlite — search

#hacktoberfest #ai #llm #go #android #linux #docker #macos #nodejs #windows #security #pytorch #ios #openai

(★ 79)

Code for "6D Object Pose Regression via Supervised Learning on Point Clouds" @ICRA2020

CloudPose 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# CloudPose
This is an tensorflow implementation of "Learning Object Pose Estimation with Point Clouds"
![](https://raw.githubusercontent.com/GeeeG/CloudPose/master/figure/System_fig.png)
# Files
1. **data**: a color, a depth and a label image for testing.
2. **models**: a python file defining model structure.
3. **object_model_tfrecord**: full object models for visualization purpose.
4. **trained_network**: a trained network.
5. **utils**: utility files for defining model structure.
6. **log**: directory to store log files during training.
7. **losses**: loss functions for training.
8. *data_process_tools.py*: tools for data pre-processing.
9. *object_6d_pose.py*: main file for testing object 6d pose estimation with a trained network.
10. *train_6d_pose.py*: script for training a network.

# Requirements
* Tensorflow-GPU (tested with 1.12.0)
* [transforms3d](https://matthew-brett.github.io/transforms3d/)
* [open3d](http://www.open3d.org/docs/getting_started.html) for visualization

# Test a trained network
1. Available classes in the test scenario are: 
 * 1: 003_cracker_box
 * 3: 005_tomato_soup_can
 * 4: 006_mustard_bottle
 * 9: 011_banana
 * 14: 035_power_drill
2. After activate tensorflow
```
python object_6d_pose.py --trained_model trained_network/20190222-130143/model.ckpt --batch_size 1 --target_class 9
```
* --trained_model: directory to trained model (*.ckpt).
* --batch_size: set to 1 because only 1 test example is provided in this test.
* --target_class: target class for pose estimation.
* Translation prediction is in unit meter.
* Rotation prediction is in axis-angle format.
3. Result
* If you turn on visualization with **b_visual=True**, you will see the following displays which are partially observed point cloud segments (colored) overlaid with object model (green) with pose estimates. 
* The coordinate is the **object coordinate**, object segment is viewed in the **camera coordinate** 
<p float="center">
  <img src="/figure/1.png" width="150" />
  <img src="/figure/3.png" width="150" /> 
  <img src="/figure/4.png" width="150" />
  <img src="/figure/9.png" width="150" />
  <img src="/figure/14.png" width="150" />
</p>

# Train a network
1. Training data in **tfrecord** format is available
* Download [zip file](https://drive.google.com/file/d/10gvvJdllvl0mMGvpiwbZ8KEzVlJS2gyZ/view?usp=sharing)
* Unzip and place all 42 files in **ycb_video_tfRecords/FPS1024/**
* Change path **data_dir** and **object_model_dir** in **train_6d_pose.py** according to your setup
2. Run script
```
python train_6d_pose.py
```
3. Log files and trained model is store in **log**
4. For reproducing the results in the paper, it takes around 22 hours to trained the model for 90 epoches on a TITAN X (Pascal). Better results might be achieved if you have more time and patience for training :D

# Create your own training data
This is an example on how to create the **tfrecords** in **ycb_video_tfRecords/FPS1024/**.
1. For an easier data access, we first write the YCB_Video_Dataset in to **tfrecords**, sorted by sequence.
* In **ycb_video_data_tfRecords/script**, run script
```
1_ycb_to_tfrecord.py
```
This will create files e.g. **0000.tfrecords** (for master_chef_can) in **sort_by_seq**. An example can be downloaded [here](https://drive.google.com/file/d/1ZHT9cfa0siv3KGK-fpYhWh8X7pQVCKdU/view?usp=sharing).

2. Create point cloud segments with pose labels for training by running script
```
2_dataset_to_tfRecord_small.py
```
This will write the training data into **FPS1024**.

Note that the point cloud segment (contains 256 points) is a **small** subsample of original segments (contains ~thousand(s) of points) created with farthest point sampling (FPS) with random starting point.

# Evaluate trained network
(to be added)
 
# Citation
Further details are available in our paper on the subject. If you use this code in an academic context, please consider cite the paper:
[arxiv version](https://arxiv.org/abs/2001.08942)

# Video
[![Video](https://img.youtube.com/vi/OhNMngzxFDQ/0.jpg)](https://www.youtube.com/watch?v=OhNMngzxFDQ)

BiBTeX:
```
@inproceedings{gao2020cloudpose,
      title={6D Object Pose Regression via Supervised Learning on Point Clouds},
      author={G. Gao, M. Lauri, Y. Wang, X. Hu, J. Zhang and S. Frintrop},
      booktitle={ICRA},
      year={2020}
    }
```

# Acknowledgement
* The building block for this system is [PointNet](https://github.com/charlesq34/pointnet).

CloudPose

CloudPose 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

CloudPose

CloudPose 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README