kevinzakka / form2fit Goto Github PK

View Code? Open in Web Editor NEW

81.0 13.0 23.0 54.83 MB

[ICRA 2020] Train generalizable policies for kit assembly with self-supervised dense correspondence learning.

Home Page: https://form2fit.github.io/

License: MIT License

Python 99.48% Shell 0.52%

robotics perception deep-learning self-supervised-learning pytorch

form2fit's People

Contributors

Stargazers

Watchers

form2fit's Issues

I would like to build a learning pipeline and an inference pipeline.

I am very interested in your paper and would like to reproduce the paper and run it using the data I have prepared.
The github has the ORB-PE executable code, but I don't think it has a learning pipeline and an inference pipeline for your proposed method. If you can provide them, I would like to have them. If not, I would appreciate it if you could answer the following questions

Are the suction, placement, and correspondence modules trained separately?
Does download_weights.sh supply only the weight parameters for the correspondence module?
How can I train the suction and placement modules using the data dropped in download_data.sh?

I know it's been a while since my presentation, but I'd be happy to hear your answers.

Translated with DeepL.com (free version)

Testing on real robot

Could you share the code for controlling the UR robot?
If you need any help I will be more than happy to contribute.

Couldn't find matching branch in code

Hi, I would like to say many thanks for the works you have done. I wonder if the matching branch (show in picture below) is included in the code. I can only find the rotation matching which will output the best similar angle between a pair of pick and place, but can not find the keypoint matching part which should be used to do position alignment. If it's somewhere inside the code please kindly guide me to it. Thank you so much!

!Error: main.py "black-floss" "train"

Dear Kevin Zakka,

When I learn the form2fit project, I can not test the GUI's main.py successfully.

Cd D:\Users\pgjgg\PycharmProjects\form2fit\form2fit\form2fit\code\gui
main.py "black-floss" "train"

bug note:

(venv) D:\Users\pgjgg\PycharmProjects\form2fit\form2fit\form2fit\code\gui>main.py "black-floss" "train"
Traceback (most recent call last):
  File "D:\Users\pgjgg\PycharmProjects\form2fit\form2fit\form2fit\code\gui\main.py", line 422, in <module>
    window = Debugger(args)
  File "D:\Users\pgjgg\PycharmProjects\form2fit\form2fit\form2fit\code\gui\main.py", line 43, in __init__
    self._init_loader_and_network()
  File "D:\Users\pgjgg\PycharmProjects\form2fit\form2fit\form2fit\code\gui\main.py", line 62, in _init_loader_and_network
    num_channels=self._num_channels,
  File "d:\users\pgjgg\pycharmprojects\form2fit\form2fit\form2fit\code\ml\dataloader\correspondence.py", line 683, in get_corr_loader
    num_channels,
  File "d:\users\pgjgg\pycharmprojects\form2fit\form2fit\form2fit\code\ml\dataloader\correspondence.py", line 62, in __init__
    self._get_filenames()
  File "d:\users\pgjgg\pycharmprojects\form2fit\form2fit\form2fit\code\ml\dataloader\correspondence.py", line 85, in _get_filenames
    self._filenames.sort(key=lambda x: int(x.split("/")[-2]))
  File "d:\users\pgjgg\pycharmprojects\form2fit\form2fit\form2fit\code\ml\dataloader\correspondence.py", line 85, in <lambda>
    self._filenames.sort(key=lambda x: int(x.split("/")[-2]))
ValueError: invalid literal for int() with base 10: 'data'

I also try to adjust the environment,just like as:

But it didn't work. And I make a issue about torch.

Hope your help,Thanks.

Can't set it up.

I have read your paper and would like to reproduce it in action.

However, I can't install 'walle' in the setup procedure.

Is there an environment, such as Docker, that I can use to easily run it?

If you have a Dockerfile, please share it with me.

!Error: python auc.py ORB-PE_acc.pkl --debug=True

Dear Kevin Zakka,

When I learn the form2fit project,I can test these code successfully.

Cd D:\Users\pgjgg\PycharmProjects\form2fit\form2fit\form2fit\code\baseline\
python eval_baseline.py --debug=True	 
Cd D:\Users\pgjgg\PycharmProjects\form2fit\form2fit\form2fit\benchmark\
python eval.py ORB-PE_poses.pkl  --debug=True

But,I can not test auc.py successfully in the way.

Cd D:\Users\pgjgg\PycharmProjects\form2fit\form2fit\form2fit\benchmark\
python auc.py ORB-PE_acc.pkl --debug=True

bug note:

(venv) D:\Users\pgjgg\PycharmProjects\form2fit\form2fit\form2fit\benchmark>python auc.py ORB-PE_acc.pkl --debug =true
D:\Users\pgjgg\PycharmProjects\form2fit\venv\lib\site-packages\numpy\core\fromnumeric.py:3373: RuntimeWarning: Mean of empty slice.
  out=out, **kwargs)
D:\Users\pgjgg\PycharmProjects\form2fit\venv\lib\site-packages\numpy\core\_methods.py:170: RuntimeWarning: invalid value encountered in double_scalars
  ret = ret.dtype.type(ret / rcount)
Traceback (most recent call last):
  File "auc.py", line 99, in <module>
    aucs[metric_name[4:]][method_name] = np.trapz(accs, thresholds)
  File "<__array_function__ internals>", line 6, in trapz
  File "D:\Users\pgjgg\PycharmProjects\form2fit\venv\lib\site-packages\numpy\lib\function_base.py", line 4078, in trapz
    shape[axis] = d.shape[0]
IndexError: list assignment index out of range

Hope your help,Thanks.

kevinzakka / form2fit Goto Github PK

form2fit's People

Contributors

Stargazers

Watchers

Forkers

form2fit's Issues

I would like to build a learning pipeline and an inference pipeline.

Testing on real robot

Couldn't find matching branch in code

!Error: main.py "black-floss" "train"

Can't set it up.

!Error: python auc.py ORB-PE_acc.pkl --debug=True

Recommend Projects

React

Vue.js

Typescript

TensorFlow

Django

Laravel

D3

Recommend Topics

javascript

web

server

Machine learning

Visualization

Game

Recommend Org

Facebook

Microsoft

Google

Alibaba

D3

Tencent