HumanML3D Preparation Guide (Unofficial)

[Zeyi-Lin]

🎉I have successfully run through the training process of MotionLCM, this is a great job!

For preparing the HumanML3D dataset, two steps need to be done:

1. Download all files from the Google Drive to the datasets/humanml3d directory, which include train.txt, val.txt, test.txt, train_val.txt, all.txt, Std.npy, Mean.npy and new_joint_vecs.rar. Unrar the new_joins_vecs.rar.

2. texts.zip: Download it from here and unzip it in the datasets/humanml3d directory.

[LinghaoChan]

Thanks for the suggestion. The guidances have been updated in the readme.

[robbinhan]

🎉I have successfully run through the training process of MotionLCM, this is a great job!

In preparing the training dataset, two steps need to be taken:

1. Download all files from this Google Drive to the datasets/humanml3d directory.
2. Unrar the three compressed files new_joins.rar, new_joins_vecs.rar, and texts.rar, and the final directory structure should be as shown in the README.

I did not find the texts.rar file.

[Zeyi-Lin]

🎉I have successfully run through the training process of MotionLCM, this is a great job!
In preparing the training dataset, two steps need to be taken:

1. Download all files from this Google Drive to the datasets/humanml3d directory.
2. Unrar the three compressed files new_joins.rar, new_joins_vecs.rar, and texts.rar, and the final directory structure should be as shown in the README.

I did not find the texts.rar file.

texts.zip download at https://github.com/EricGuo5513/HumanML3D/tree/main/HumanML3D

[robbinhan]

@Zeyi-Lin Thank you very much

[Zeyi-Lin]

@Zeyi-Lin Thank you very much

I’m happy to help. Good luck.

[zymcloud]

I am sorry, I meet the problem that I don't have permission to access this link Google Drive.