Human3.6M preprocessing scripts

These scripts help preprocess Human3.6M dataset so that it can be used with class Human36MMultiViewDataset.

Here is how we do it (brace yourselves):

0. Make sure you have a lot (around 200 GiB?) of free disk space. Otherwise, be prepared to always carefully delete intermediate files (e.g. after you extract movies, delete the archives).

1. Please do not ask me for a copy of the Human3.6M dataset. I do not own the data, nor do I have permission to redistribute it. You will need to create an account at http://vision.imar.ro/human3.6m/ to download the dataset and put the raw data under $SOMEWHERE_ELSE/h36m_raw_data

2. Choose a folder for the processed data to be stored. Make it accessible like $TransPose_ROOT/data/human36m where ROOT_PATH is data/human36m by default. Modify RAW_DATA_PATH to be $SOMEWHERE_ELSE/h36m_raw_data, and NUM_PROCESS to be the one your environment allows in the first two lines in process_h36m.sh.

3. Download h36m.zip (provided by Julieta Martinez) from Google Drive

   mkdir -p $TransPose_ROOT/data/human36m/extra/una-dinosauria-data
   

   and put the zipfile under $TransPose_ROOT/data/human36m/extra/una-dinosauria-data.

4. We have integrated all the processing codes into process_h36m.sh and you can simply run it under $TransPose_ROOT to finish all the procedures instructed by mvn/datasets/human36m_preprocessing/README.md:

   cd $TransPose_ROOT
   # in our environment, this took around 5 hours with 40 processes
   sh process_h36m.sh
   

5. If everything goes well, your data directory tree will be like this (our code also support read .zip data, you can simply zip the processed folder to be processed.zip, and change the data_format in the config file to be 'zip'.):

   $TransPose_ROOT
   |-- data
   `-- |-- human36m
       `-- |-- extra
           |   |-- una-dinosauria-data
           |   `-- |-- h36m.zip 
           |   |   |-- h36m   
           |   |   `-- |-- S1
           |   |       |-- S11
           |   |       |-- S5
           |   |       |-- S6
           |   |       |-- S7
           |   |       |-- S8
           |   |       |-- S9
           |   |       `-- cameras.h5
           |   |-- bboxes-Human36M-GT.npy
           |   |-- human36m-multiview-labels-GTbboxes.npy
           |   `-- mean_and_std_limb_length.h5
           |-- extracted
           |   |-- S1
           |   `-- |-- Poses_D2_Positions
           |   |   |-- Poses_D3_Positions
           |   |   |-- Poses_D3_Positions_mono
           |   |   |-- Poses_D3_Positions_mono_universal
           |   |   `-- Videos
           |   |-- S11
           |   |-- S5
           |   |-- S6
           |   |-- S7
           |   |-- S8
           |   `-- S9
           `-- processed (or processed.zip)
               |-- S1
               `-- |-- Directions-1
               |   `-- |-- imageSequence
               |   |   |-- imageSequence-undistorted
               |   |   `-- |-- 54138969
               |   |   |   `-- |-- img_000001.jpg      
               |   |   |   |-- |-- img_000008.jpg      
               |   |   |   |-- |-- img_000011.jpg      
               |   |   |   `-- |-- ...
               |   |   |   |-- 55011271
               |   |   |   |-- 58860488
               |   |   |   `-- 60457274
               |   |   `-- annot.h5
               |   |-- Directions-2
               |   |-- Discussion-1
               |   `-- ...
               |-- S11
               |-- S5
               |-- S6
               |-- S7
               |-- S8
               `-- S9    
   

6. [If you want to learn more for processing the data] All the detailed instructions in process_h36m.sh can be found at mvn/datasets/human36m_preprocessing/README.md.