Different number of files after running the Data Preparation Script

lisiyao21 / Bailando

Code for CVPR 2022 paper "Bailando: 3D dance generation via Actor-Critic GPT with Choreographic Memory"

Other

381 stars 59 forks source link

Different number of files after running the Data Preparation Script #51

Open XulongT opened 3 months ago

XulongT commented 3 months ago

When following the steps outlined in the documentation to prepare the AIST++ dataset, after downloading the annotations and unzipping them into the './aist_plusplus_final/' folder, and then running the './prepare_aistpp_data.sh' script, the resulting number of files in the 'data/aistpp_train_wav' directory is 980. However, your paper mentions 952 files intended for training. This discrepancy requires clarification and investigation.

Sun-Happy-YKX commented 3 months ago

I've met the same question. @lisiyao21 could you please give me a hand.

Sun-Happy-YKX commented 3 months ago

980 is found in AIST++ original dataset split file "crossmodal_train.txt"

applesleam commented 1 month ago

There is a ignore list: https://github.com/lisiyao21/Bailando/blob/main/ignore_list.txt to delete the low quality motion and result in 952 files eventually.