-
|
Can anyone show a screenshot of what is supposed to happen when you import a csv file of already labelled deeplabcut data into SLEAP? Does it make the training faster? What are the steps you still have to do after? |
Beta Was this translation helpful? Give feedback.
Replies: 5 comments 7 replies
-
|
Hi @jannhan, Apologies for the delay! We're a bit behind on support responses at the moment. You should be able to import training data (manual labels, not predictions) from the File menu. It should not require any additional steps -- you'll just see the labeled frames imported right away. This feature allows you to reuse annotations that you've already generated, saving you time from having to start from scratch. Did you run into issues when using this? If nothing happened in the GUI, did you get any error messages in the terminal? Cheers, Talmo |
Beta Was this translation helpful? Give feedback.
-
|
No problem, I totally understand! The first error/warning message that came up said that I had to install libnvinfer.so <http://libnvinfer.so/>.6 and libnvinfe_plugin.so <http://libnvinfe_plugin.so/>.6, and this is after I’ve already downloaded TensorRT, TensorFlow, cuDNN, CUDA, so I’m still stuck. Please let me know if you know what could be going on!
Thank you so much for your time and help!
Janna
… On Jul 19, 2024, at 7:59 PM, Talmo Pereira ***@***.***> wrote:
Hi @jannhan <https://github.com/jannhan>,
Apologies for the delay! We're a bit behind on support responses at the moment.
You should be able to import training data (manual labels, not predictions) from the File menu.
It should not require any additional steps -- you'll just see the labeled frames imported right away.
This feature allows you to reuse annotations that you've already generated, saving you time from having to start from scratch.
Did you run into issues when using this? If nothing happened in the GUI, did you get any error messages in the terminal?
Cheers,
Talmo
—
Reply to this email directly, view it on GitHub <#1833 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BHT2WUU5RD76AJ2GWJ454GDZNGR4VAVCNFSM6AAAAABJ6ZRJZSVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTAMBZHE4DOOI>.
You are receiving this because you were mentioned.
|
Beta Was this translation helpful? Give feedback.
-
|
Hi @jannhan, This error can be safely ignored and is just related to GPU dependencies. I'd advise that you install SLEAP via the recommended conda package method, which will install it in an isolated conda environment where it handles GPU dependencies and won't interact with other packages that you have installed. |
Beta Was this translation helpful? Give feedback.
-
|
If I ignore this warning message, and import the DeepLabCut data, an error
comes up that says "ValueError: Skeleton already has node named (bodypart1)"
…On Sun, Jul 28, 2024 at 11:10 PM Janna Han ***@***.***> wrote:
No problem, I totally understand! The first error/warning message that
came up said that I had to install libnvinfer.so.6 and libnvinfe_plugin.so.6,
and this is after I’ve already downloaded TensorRT, TensorFlow, cuDNN,
CUDA, so I’m still stuck. Please let me know if you know what could be
going on!
Thank you so much for your time and help!
Janna
On Jul 19, 2024, at 7:59 PM, Talmo Pereira ***@***.***>
wrote:
Hi @jannhan <https://github.com/jannhan>,
Apologies for the delay! We're a bit behind on support responses at the
moment.
You should be able to import training data (manual labels, not
predictions) from the File menu.
It should not require any additional steps -- you'll just see the labeled
frames imported right away.
This feature allows you to reuse annotations that you've already
generated, saving you time from having to start from scratch.
Did you run into issues when using this? If nothing happened in the GUI,
did you get any error messages in the terminal?
Cheers,
Talmo
—
Reply to this email directly, view it on GitHub
<#1833 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/BHT2WUU5RD76AJ2GWJ454GDZNGR4VAVCNFSM6AAAAABJ6ZRJZSVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTAMBZHE4DOOI>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
Beta Was this translation helpful? Give feedback.
-
|
That probably shouldn't be happening -- are you importing it into a blank project or a pre-existing one? What happens if you try just creating a brand new project and then importing the CSV? |
Beta Was this translation helpful? Give feedback.
-
|
I reinstalled SLEAP with the conda package method and the warning went
away! However, I tried opening a new project then importing the csv and the
same error comes up. What is it supposed to look like when I open it?
…On Mon, Jul 29, 2024 at 4:43 PM Talmo Pereira ***@***.***> wrote:
That probably shouldn't be happening -- are you importing it into a blank
project or a pre-existing one?
What happens if you try just creating a brand new project and then
importing the CSV?
—
Reply to this email directly, view it on GitHub
<#1833 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/BHT2WUQ7EDIOK5ZWMZSVI43ZO2SQPAVCNFSM6AAAAABJ6ZRJZSVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTAMJYGM3TINQ>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
Beta Was this translation helpful? Give feedback.
-
|
Hmm, it's supposed to show the imported data, so perhaps something funky is going on. Do you mind trying out the v1.4.1a2 prerelease? I think there might be a fix related to this issue in there. If it still doesn't work, we'll ask you for your CSV file so we can see what's going on. |
Beta Was this translation helpful? Give feedback.
-
|
Unfortunately the same error still comes up. Here is the CSV file I have
been testing this with
…On Mon, Jul 29, 2024 at 5:15 PM Talmo Pereira ***@***.***> wrote:
Hmm, it's supposed to show the imported data, so perhaps something funky
is going on.
Do you mind trying out the v1.4.1a2 prerelease
<https://github.com/talmolab/sleap/releases/tag/v1.4.1a2>? I think there
might be a fix related to this issue in there.
If it still doesn't work, we'll ask you for your CSV file so we can see
what's going on.
—
Reply to this email directly, view it on GitHub
<#1833 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/BHT2WUUM7Z3V5EOKAVZDTMDZO2WFNAVCNFSM6AAAAABJ6ZRJZSVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTAMJYGQYDAOA>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
Beta Was this translation helpful? Give feedback.
-
|
Hi @jannhan, I think you probably tried to attach the file to an email -- I think GitHub doesn't allow this. Do you mind uploading it to Google Drive or something? You might also be able to attach it from the GitHub website for this discussion --> #1833 Cheers, Talmo |
Beta Was this translation helpful? Give feedback.
-
|
Sure, let me know if this works |
Beta Was this translation helpful? Give feedback.
-
|
Hi @jannhan, Thanks for sharing the CSV! It looks like these are the predictions, not the manually labeled annotations. We could probably add support for importing predictions, but one issue is that this CSV doesn't have paths to the images or the video that the predictions correspond to. If you have the CSV with the labels that were used to train the model, that should work :) Cheers, Talmo |
Beta Was this translation helpful? Give feedback.
-
|
Thank you so so much, the manually labelled csv works! |
Beta Was this translation helpful? Give feedback.
Hi @jannhan,
Thanks for sharing the CSV! It looks like these are the predictions, not the manually labeled annotations.
We could probably add support for importing predictions, but one issue is that this CSV doesn't have paths to the images or the video that the predictions correspond to.
If you have the CSV with the labels that were used to train the model, that should work :)
Cheers,
Talmo