Request for Training Code for Reproducing VSR (Lip Reading)

[HarukiK2]

Hello,

Thank you very much for your great work and for releasing CoGenAV — I believe it is a very promising project.

I'm currently working on reproducing visual speech recognition (VSR), specifically lip reading, and I’m very interested in CoGenAV's performance on the LRS2 dataset.

I’ve checked the repository and found the inference scripts (e.g., infer_avse_avss.py, infer_vsr_avsr.py), but I couldn’t find any training scripts or instructions.

Would it be possible for you to share the training code as well?

In particular, I’d appreciate any scripts or configuration files related to training on LRS2 for the VSR task.

I understand that preparing and releasing the training code may take time, but any guidance or partial resources would be greatly appreciated.

Thanks again for your valuable contribution to the community!

[OdingdongO]

Hello,

Thank you very much for your great work and for releasing CoGenAV — I believe it is a very promising project.

I'm currently working on reproducing visual speech recognition (VSR), specifically lip reading, and I’m very interested in CoGenAV's performance on the LRS2 dataset.

I’ve checked the repository and found the inference scripts (e.g., infer_avse_avss.py, infer_vsr_avsr.py), but I couldn’t find any training scripts or instructions.

Would it be possible for you to share the training code as well?

In particular, I’d appreciate any scripts or configuration files related to training on LRS2 for the VSR task.

I understand that preparing and releasing the training code may take time, but any guidance or partial resources would be greatly appreciated.

Thanks again for your valuable contribution to the community!

Thank you for your recognition of our work. As you mentioned, we have released the code and models related to inference. Once we have decomposed and reconstructed the training code to ensure clarity, we will update the repository with that information.

[HarukiK2]

Thank you for expressing your intention to make the training code available. I really appreciate your willingness to share it in the future!

[HarukiK2]

Thanks again for your reply!

As a quick follow-up — I’m aware that the LRS3 dataset is currently unavailable through the official website.
However, since LRS3 is widely used in many VSR-related studies, I would ideally like to use it as well.

With that in mind, I was wondering if you’ve tried any alternative ways to obtain it, such as directly emailing the dataset authors or reaching out through other channels.
Any insights or suggestions would be greatly appreciated.

Thanks again for your help!

[xduzhangjiayu]

Thanks again for your reply!

As a quick follow-up — I’m aware that the LRS3 dataset is currently unavailable through the official website. However, since LRS3 is widely used in many VSR-related studies, I would ideally like to use it as well.

With that in mind, I was wondering if you’ve tried any alternative ways to obtain it, such as directly emailing the dataset authors or reaching out through other channels. Any insights or suggestions would be greatly appreciated.

Thanks again for your help!

Have you solved it? I met the same issue

[HarukiK2]

Thanks again for your reply!
As a quick follow-up — I’m aware that the LRS3 dataset is currently unavailable through the official website. However, since LRS3 is widely used in many VSR-related studies, I would ideally like to use it as well.
With that in mind, I was wondering if you’ve tried any alternative ways to obtain it, such as directly emailing the dataset authors or reaching out through other channels. Any insights or suggestions would be greatly appreciated.
Thanks again for your help!

Have you solved it? I met the same issue

Actually I haven't solved it yet

[Hu-bc]

Hello,
Thank you very much for your great work and for releasing CoGenAV — I believe it is a very promising project.
I'm currently working on reproducing visual speech recognition (VSR), specifically lip reading, and I’m very interested in CoGenAV's performance on the LRS2 dataset.
I’ve checked the repository and found the inference scripts (e.g., infer_avse_avss.py, infer_vsr_avsr.py), but I couldn’t find any training scripts or instructions.
Would it be possible for you to share the training code as well?
In particular, I’d appreciate any scripts or configuration files related to training on LRS2 for the VSR task.
I understand that preparing and releasing the training code may take time, but any guidance or partial resources would be greatly appreciated.
Thanks again for your valuable contribution to the community!

Thank you for your recognition of our work. As you mentioned, we have released the code and models related to inference. Once we have decomposed and reconstructed the training code to ensure clarity, we will update the repository with that information.

Hi there! Just wanted to gently follow up on the training code release. We are really excited about this work and are eager to try training it on LRS2. Do you have a rough estimate of when the training scripts might be available? Thanks again for your efforts!