search

WangYueFt / dcp

346 stars 90 forks source link

how to test your own code #5

Closed zhangshabao closed 5 years ago

zhangshabao commented 5 years ago

Excuse me, I would like to ask you how to test your own code?

WangYueFt commented 5 years ago

Excuse me, I would like to ask you how to test your own code?

Hi,

We're happy to answer any questions. Please ask here.

Thanks, Yue

zhangshabao commented 5 years ago

Using own data for testing, is that data to be converted to a.h5 file?Can I use TXT file directly?

WangYueFt commented 5 years ago

You can just change the data loader or directly load your data, then convert it to torch tensor.

On Thu, Jun 27, 2019 at 6:13 PM zhangshabao notifications@github.com wrote:

Using own data for testing, is that data to be converted to a.h5 file?Can I use TXT file directly?

— You are receiving this because you modified the open/close state.

Reply to this email directly, view it on GitHub https://github.com/WangYueFt/dcp/issues/5?email_source=notifications&email_token=ACMGCNQOXZV7T73ZTVDPSRLP4VQTXA5CNFSM4H34QH42YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODYYY5IA#issuecomment-506564256, or mute the thread https://github.com/notifications/unsubscribe-auth/ACMGCNWELM3ANMSZBHWNO53P4VQTXANCNFSM4H34QH4Q .

zhangshabao commented 5 years ago

Does the requirement of point cloud quantity need to be controlled at 2048?

WangYueFt commented 5 years ago

No. If you first train on your data.

On Thu, Jun 27, 2019 at 7:13 PM zhangshabao notifications@github.com wrote:

Does the requirement of point cloud quantity need to be controlled at 2048?

— You are receiving this because you modified the open/close state.

Reply to this email directly, view it on GitHub https://github.com/WangYueFt/dcp/issues/5?email_source=notifications&email_token=ACMGCNU2V7FURIOMXU2OR4DP4VXTLA5CNFSM4H34QH42YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODYY3HPQ#issuecomment-506573758, or mute the thread https://github.com/notifications/unsubscribe-auth/ACMGCNW5BK5U6YUD5DK27X3P4VXTLANCNFSM4H34QH4Q .

zhangshabao commented 5 years ago

If I've trained with your data,should the data now be tested in a format consistent with your data?Thank you for your answer！

BlankCheng commented 5 years ago

The author has provided download function, it will download modelnet40 automatically. If you want to train with your own data, just change load_data in data.py to fit~

zhangshabao commented 5 years ago

The author has provided download function, it will download modelnet40 automatically. If you want to train with your own data, just change load_data in data.py to fit~

Can you share the code?Some questions are not very clear at the beginning. thank you !

BlankCheng commented 5 years ago

The author has provided download function, it will download modelnet40 automatically. If you want to train with your own data, just change load_data in data.py to fit~

Can you share the code?Some questions are not very clear at the beginning. thank you !

ummm I have not trained it and changed the code, just did some inference on ycb dataset using pretrained model given by author.

zhangshabao commented 5 years ago

The author has provided download function, it will download modelnet40 automatically. If you want to train with your own data, just change load_data in data.py to fit~

Can you share the code?Some questions are not very clear at the beginning. thank you !

ummm I have not trained it and changed the code, just did some inference on ycb dataset using pretrained model given by author.

I've trained with the author's data set, and I think my data should be converted to match it.Confusion about what that data set contains.

BlankCheng commented 5 years ago

The author has provided download function, it will download modelnet40 automatically. If you want to train with your own data, just change load_data in data.py to fit~

Can you share the code?Some questions are not very clear at the beginning. thank you !

ummm I have not trained it and changed the code, just did some inference on ycb dataset using pretrained model given by author.

I've trained with the author's data set, and I think my data should be converted to match it.Confusion about what that data set contains.

http://modelnet.cs.princeton.edu/ Modelnet40 is described in this link. "data" in load_data is CAD point cloud , maybe a N * 3 matrix, "label" is its category, a number from 0 to 39. If you want to train the classification with your own data, just fit in this format of "data" and "label", no need to save it in h5.

zhangshabao commented 5 years ago

The author has provided download function, it will download modelnet40 automatically. If you want to train with your own data, just change load_data in data.py to fit~

Can you share the code?Some questions are not very clear at the beginning. thank you !

ummm I have not trained it and changed the code, just did some inference on ycb dataset using pretrained model given by author.

I've trained with the author's data set, and I think my data should be converted to match it.Confusion about what that data set contains.

http://modelnet.cs.princeton.edu/ Modelnet40 is described in this link. "data" in load_data is CAD point cloud , maybe a N * 3 matrix, "label" is its category, a number from 0 to 39. If you want to train the classification with your own data, just fit in this format of "data" and "label", no need to save it in h5.

Is it convenient to leave your email?Can we continue our conversation?My QQ is 1047253497.Thank you~